1
Objective or Subjective –
Anti-treaty shopping policy in select Asian jurisdictions in the post-BEPS
world
Sunghak (“Andy”) Baik / Matthias Petutschnig1
ABSTRACT
BEPS Action 6 proposes two distinct anti abuse measures to be incorporated into the
OECD Model Convention and subsequently into the various bilateral tax treaties: A
Limitation on Benefits (LoB) clause and Principal Purpose Test (PPT). While both anti
abuse measures are new to the OECD Model Convention, various countries around the
world have implemented either LoB or PPT clauses or both into their tax treaties. This
paper analyses the treaty network of eight Asian / Pacific jurisdictions (Australia, China,
Hong Kong, Japan, Malaysia, Singapore, and Taiwan) with respect to the anti-abuse
measures employed in these treaties. The majority of the more than 500 treaties in
the sample do not included an anti-abuse measure of any kind. While the use of anti-
abuse rules in general is highly diverse, the choice of the preferred measure if an anti-
abuse rule is incorporated in the treaty is quite homogeneous. The one measure most
often used is the principal purpose test (or a variation thereof), with 113 individual
treaties containing that test. LoB clauses are used in only 16 treaties. The historical
development of the treaty networks shows a strong increase in the (relative)
importance of the PPT since 43% of all new treaties concluded after 2009 contain such
a provision while only 5% of these treaties contain a LoB.
Sunghak (“Andy”) Baik is a Principal at Ernst & Young LLP’s Financial Services practice and is based in San Francisco. Andy regularly advises private equity funds, sovereign wealth funds and others in the financial services sector on international tax matters. Prior to his recent repatriation to the US, he spent 11 years in Singapore and Korea advising on both Asia inbound and Asia outbound tax matters.
Dr Matthias Petutschnig is Assistant Professor of Taxation at WU Vienna University of Economics and Business – Department for Finance, Accounting and Statistics (Tax Management Group).
1 The authors would like to acknowledge the invaluable assistance provided by Vivienne Junzhao Ong and Daniela Arth on this paper.
2
1. Introduction
Since 2008, scientific literature2 and subsequently media reports have drawn attention to the fact
that some highly profitable multinational companies seem to pay comparatively little to no corporate
income tax especially in the source country. The effective tax rates on foreign profits of, for example,
Google and Apple were reported to be 3% and 1%, respectively.3 The fact the some multinationals
are able to considerably reduce their tax burden by exploiting national differences and loopholes in
existing tax rules, often described as base erosion and profit shifting (BEPS), suggests that the
taxation of multinational firms is in need of reform.4 The necessity for reform is reflected by the
intense public debate surrounding profit shifting and tax avoidance by multinational firms. Given that
many countries face high levels of public debt and strong pressure to generate (additional) tax
revenue, it is not surprising that this debate has brought the taxation of multinationals to the top of
the international political agenda.
In 2013 the Organization for Economic Co-operation and Development (OECD) published a global
action plan comprising 15 actions aimed at tackling base erosion and profit shifting of multinational
enterprises (“BEPS Action Plan”).5 The BEPS Action Plan suggests a variety of legislative and
administrative measures which aim at eliminating double non-taxation and under-taxation (i.e., a
taxation level which is perceived as too low). After two years of intense work within the OECD and
after numerous public consultations, the final reports on the actions were delivered on 5 October
20156 and endorsed by the G20 in February 2016.7
The OECD Action Plan and the final reports include a broad array of recommendations on domestic
and/or bilateral measures aimed at preventing base erosion and profit shifting in the future. Existing
domestic and international tax rules should be modified in order to more closely align the allocation
of income with the economic activity that generates that income. The BEPS Action Plan identifies
treaty abuse, and in particular treaty shopping, as one of the most important sources of BEPS
concerns. The final report on Action 6, titled Preventing the Granting of Treaty Benefits in
Inappropriate Circumstances (“Action 6”), contains a comprehensive rule to limit or deny the benefits
2 See for an extensive literature review D. Dharmapala, "What Do We Know About Base Erosion and Profit Shifting? A Review of the Empirical Literature" (2014), 35 Fiscal Studies 12, 421–448.
3 See note 2.
4 See C. Fuest, C. Spengel, K. Finke, J.H. Heckemeyer and H. Nusser, "Profit Shifting and “Aggressive” Tax Planning by Multinational Firms: Issues and Options for Reform" (2013) 5 World Tax Journal 3, 307-324.
5 OECD (2013). Action Plan on Base Erosion and Profit Shifting. OECD.
6 OECD (2015). BEPS Final Reports. Available at: http://www.oecd.org/ctp/beps-2015-final-reports.htm.
7 G20 (2016). Communiqué G20 Finance Ministers and Central Bank Governors Meeting. G20 Summit China. Available at: http://www.g20.org/English/Documents/Current/201603/t20160302_2182.html (25 Feb 2017).
3
of a tax treaty when the treaty is purposefully used to attain unduly benefits. The proposed rule
consists of an objective (“Limitation on Benefits clause” – LoB) and a rather subjective (“Principal
Purpose Test” – PPT) anti-abuse measure. Although the BEPS Action Plan leaves some flexibility to
the adopting countries, it expects that countries achieve a minimum standard of protection against
treaty abuse. This could be achieved through either the combined inclusion of the LOB clause and the
PPT rule or the inclusion of either the LOB clause or the PPT rule.8
This paper analyzes the proposal made in Action 6 against the backdrop of the historical evolution
and the current state of bilateral anti treaty abuse measures. With a special focus on eight
Asian/Pacific jurisdictions9 and their treaty networks, the paper aims at providing guidance to
domestic policy makers and bilateral treaty negotiators as to which proposed measure might be
preferable to counter tax treaty abuse in the future.
The remainder of this paper is structured as follows: First we discuss the final report of Action 6 and
the proposed measures therein in section 2. Section 3 addresses the historical evolution of anti-
treaty-abuse measures while section 4 presents the current state of anti-treaty-abuse measures in
the tax treaty networks of the eight jurisdictions covered. Building on these sections we present our
conclusions, policy implications and recommendations in section 5.
2. BEPS - Action 6
The BEPS Action Plan identifies treaty abuse, and in particular treaty shopping, as one of the most
important sources of BEPS concerns.10 Action 6 (“Prevent Treaty Abuse“) is specifically aimed at
tackling this issue. The final report of Action 6 was published in the first set of deliverables in
September 2014. The final report addresses three areas of the OECD’s work on treaty abuse:
a) A clarification that tax treaties are not intended to be used to generate double non-taxation.
This clarification will inter alia change the official title of the OECD Model Treaty in a way that
it covers the objective to avoid not only double-taxation but also double non-taxation.
b) A set of Anti-Treaty Abuse provisions and/or domestic rules to prevent treaty abuse.
c) Tax policy considerations that countries should consider before entering into a tax treaty.
8 See L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse" (2014), 43 Intertax 2, 122-146.
9 Australia, China, Hong Kong, Japan, Malaysia, Singapore, South Korea, and Taiwan.
10 See BEPS Action 6 Introduction.
4
While Action 6 as well as the whole BEPS Action Plan seems to have a greater chance of being
implemented than prior OECD initiatives on treaty abuse, tackling treaty abuse is, however, not a
novelty. The conceptual foundations of Action 6 have been laid out already decades ago. Although
the OECD’s work on treaty abuse started to develop in 1977,11 it was particularly the 2003 revision to
the OECD Commentary on Article 1 of the OECD Model Convention (OECD-MC) which offered tax
authorities with ever stronger measures to counter treaty abuse.
From 1977 until the 2003 revision, the Commentary emphasized that the purpose of tax treaties was
to foster international trade and investment by eliminating international double taxation. It
nonetheless added that tax treaties “should not, however, help tax avoidance or evasion.”12 The
responsibility though was put on national legislators to enact domestic anti-avoidance rules to
counter the exploitation of differences in domestic tax legislations. The 1977 Commentary
recognized that domestic anti-avoidance measures could conflict with the provisions of tax treaties
(treaty override) and thus provided that countries which had enacted domestic anti-avoidance rules
may aim to “preserve the application of provisions of this kind” in their tax treaties.13
The 2003 revision of the Commentary on Article 1 OECD MC saw a fundamentally change in the view
of the OECD on the improper use of tax treaties and the relationship between domestic anti-
avoidance rules and tax treaties. The Commentary for the first time presented the prevention of tax
avoidance or evasion as a self-standing14 – albeit ancillary – purpose of tax treaties.15 Additionally,
the 2003 Commentary clarified that general anti-avoidance rules (“GAARs”) and/or judicial doctrines
(such as substance-over-form, economic substance) are part of the basic rules for determining the
tax liability.16 Such rules are not addressed in tax treaties and are neither affected by them nor
considered a treaty override. The 2003 Commentary therefore contends that, as a general rule, there
is no conflict between GAARs and judicial doctrines on one hand and tax treaties on the other.17
11 For a comprehensive historical overview of the Commentary on treaty abuse, see L. De Broe, International Tax Planning and Prevention of Abuse, IBFD, 2008, 377–386.
12 See OECD Commentary (1977), Art. 1, 7.
13 See L. De Broe, International Tax Planning and Prevention of Abuse, Amsterdam, IBFD, 2008, 377 et seq; L De Broe and J Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146.
14 The ‘principal’ purpose of tax treaties remained the prevention of double taxation, see B.J. Arnold, “Tax Treaties and Tax Avoidance: The 2003 Revisions to the Commentary to the OECD Model” (2004), 58 Bulletin for International Taxation 6, 248.
15 Including the prevention of treaty abuse as one of the purposes of a tax treaty would according to Article 31(1) of the Vienna Convention on the Law of Treaties (VCLT) require a treaty interpreter to take account of this object and purpose, as a consequence of which tax treaties might be interpreted to deny treaty benefits to abusive transactions (see B.J. Arnold and S. van Weeghel, “The Relationship between Tax Treaties and Domestic Anti-Abuse Measures”, in G. Maisto (Ed.), Tax Treaties and Domestic Law, Amsterdam, IBFD, 2006, 90; L De Broe and J Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146).
16 See OECD Commentary (2003/2014), Art. 1, s. 9.2 and s. 22.1.
17 See OECD Commentary (2003/2014), Art. 1, s. 9.2 and s. 22.1.
5
Even though the 2003 Commentary did not provide a definition of what constitutes an abuse of tax
treaties, it offers a ‘guiding principle’, which provides that the benefits of a double taxation
convention should not be available when two elements, a subjective and an objective element, are
present:
a) “a main purpose for entering into certain transactions or arrangements was to secure a more
favourable tax position” (subjective element); and
b) “obtaining that more favourable treatment in these circumstances would be contrary to the
object and purpose of the relevant provisions” (objective element).
However, it is also noted that “it should not be lightly assumed that a taxpayer is entering
[purposefully] into [this] type of abusive transactions”.18
The 2003 Commentary points out, that the fact that domestic (general) anti-abuse rules might apply
to deny treaty benefits does not imply that (specific) treaty-based anti-abuse rules aimed at
preventing particular forms of tax avoidance are unnecessary.19 In that respect, the 2003
Commentary suggests a range of provisions (such as look through approach; subject-to-tax-clauses;
limitation-on-benefits provisions) that treaty negotiators might consider.20
The 2003 Commentary was ambiguously received by scholars, tax administrations and tax
practitioners. While it was well received by certain scholars,21 it was widely criticized on various
aspects by others.22 More importantly, various OECD Member States have made observations to the
Commentary on Article 1 after the 2003 revision.23 Additionally, while scholars repeatedly refer to
the significance and relevance of the Commentary of the OECD-MC while interpreting specific
bilateral treaties modelled after the OECD-MC,24 the actual relevance of the Commentary is far more
18 See OECD Commentary (2003/2014), Art. 1, s. 9.5.
19 OECD Commentary (2003/2014), Art. 1, s. 9.6.
20 OECD Commentary (2003/2014), Art. 1, s. 12–21.5.
21 See J. Sasseville, “A Tax Treaty Perspective: Special Issues” in G. Maisto (Ed.), Tax Treaties and Domestic Law, Amsterdam, IBFD, 2006, 55 et seq.
22 For example L. De Broe, International Tax Planning and Prevention of Abuse, Amsterdam, IBFD, 2008, 386–404; B.J. Arnold and S. van Weeghel, ‘The Relationship between Tax Treaties and Domestic Anti-Abuse Measures’, in G. Maisto (Ed.), Tax Treaties and Domestic Law, Amsterdam, IBFD, 2006, 89 et seq.; A.J.M. Jiménez, ‘The 2003 Revision of the OECD Commentaries on the Improper Use of Tax Treaties: A Case for the Declining Effect of the OECD Commentaries?’ (2004), 58 Bulletin for International Taxation 1, 17–18; B.J. Arnold, ‘Tax Treaties and Tax Avoidance: The 2003 Revisions to the Commentary to the OECD Model’ (2004), 58 Bulletin for International Taxation 6, 244; J.J. Zornoza Pérez and A. Báez, “The 2003 Revisions to the Commentary to the OECD Model on Tax Treaties and GAARs: A Mistaken Starting Point” in M. Lang et al. (Eds.), Tax Treaties: Building Bridges between Law and Economics, (Amsterdam: IBFD, 2010), 130.
23 Ireland, Luxembourg, the Netherlands and Switzerland made observations on the Commentary position to the relationship between tax treaties and domestic anti-abuse rules in general (see OECD Commentary, Art. 1, s. 27.4–27.9).
24 Vogel argues – with reference to the High Court of Australia – that the Commentary on the OECD-MC is a guide to the current usage of terms by the parties to the treaty and thus an important source for interpretation of tax treaties (see K. Vogel, “Soft Law und Doppelbesteuerungsabkommen“, in M. Lang, J. Schuch and C. Staringer (eds.), Soft Law in der Praxis (Vienna: Linde, 2005), 145-148).
6
limited.25 Thus a ‘soft-law’ measure such as amending the OECD Commentary can by far not achieve
the same overall guidance and relevance as a ‘hard-law’ measure could achieve. The fact that the
issue of treaty abuse, and in particular treaty shopping, has been identified by the OECD as one of
the most important sources of BEPS implies that the 2003 revisions were unsatisfactory or that many
states did not execute the OECD recommendations proposed therein.
Action 6 now builds on the fundamental groundwork laid out in the 2003 revisions of the OECD
Commentary but provides the 2003 amendments with a different and enhanced legal quality. The
final report of Action 6 contains a specific anti-treaty abuse rule, which will be added to the OECD-
MC. Action 6 essentially proposes to migrate the 2003 amendments to the Commentary into the
Model Treaty itself. This proposed Article X (“Entitlement to Benefits”) includes a rather objective
anti-treaty abuse measure (Limitation-on-Benefits clause – Art X (1)-(5)) and a subjective anti-treaty
abuse measure (Principal Purpose Test – Art X (7)). Subsequently all OECD Member States, the non-
OECD G20 states as well as all other countries which commit themselves to adhere to the principles
laid out in the OEDC BEPS Action Plan are at least morally obliged to include an anti-treaty abuse
measure in their tax treaties. This could be achieved through either the combined inclusion of the
LOB clause and the PPT rule or the inclusion of either the LOB clause or the PPT rule.26
2.1. Limitation on Benefits Clause
Action 6 proposes to include a limitation-on-benefits (“LoB”) clause in the OECD-MC.27 Using LoB
clauses is an approach that has long been pioneered by the United States,28 but is a concept which is
neither foreign to the OECD member states as briefly discussed above, nor to other countries. In
general, a LoB clause is intended to provide a more objective approach to ascertaining one’s
qualification for treaty benefits (Article 1 OECD MC).29 By including a LoB clause, the contracting
25 Mössner exemplifies the legal uncertainty that is connected to the Commentary showing that the German Federal Tax Court references the Commentary only when the Commentary supports the Court’s opinion, while completely ignoring the Commentary in all other cases (see J.M. Mössner, “Diskussion zu Klaus Vogel Soft Law und Doppelbesteuerungsabkommen“, in M. Lang, J. Schuch and C. Staringer (eds.), Soft Law in der Praxis (Vienna: Linde, 2005), 149 et seq.
26 See L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse” (2015), 60 British Tax Review 3, 408-416.
27 New Art. X (1) to (6) OECD MC.
28 Compare Art. 22 of the US Model Convention.
29 See L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; R.L. Doernberg and K. van Raad, “The 1996 United States Model Income Tax Convention. Analysis, Commentary and Comparison”, The Hague, Kluwer Law International, 1997, 172; F.A. Vega Borrego, “Limitation on Benefits Clauses in Double Taxation Conventions”, The Hague, Kluwer Law International, 2006, 92–93 and 115.
7
states express their wish to only grant treaty protection to taxpayers that, in addition to being
residents, either
a) carry out real business activities,
b) have a sufficient nexus to their residence state; or
c) have bona fide motives.
These prerequisites are evaluated using a series of alternative tests. Persons that are residents of one
of the contracting states need to satisfy at least one of those tests in order to be entitled to certain,
or all, treaty benefits.30 However, only companies or other legal entities are assessed against these
alternative tests. Individuals and governmental entities are deemed to regularly have bona fide
motives per se, and treaty abuses by this heterogeneous group are seen as less common and thus
not warranting filtration by a LoB test.31
Because of its mechanical, rule-based nature, a LoB clause is considered to imply a greater amount of
legal certainty in granting treaty benefits as compared to more subjective approaches, such as the
PPT rule (see below).32 However, this benefit may also be one of its greatest drawbacks. Capturing
situations of deemed treaty abuse with a number of mechanical tests necessarily implies that the LoB
clause becomes an “awesomely complex and dense construct“.33
Article X Paragraph 1 provides the general rule that a resident of a contracting state will only be
entitled to a treaty benefit either if the resident is a qualified person (as defined in paragraph 2) or
unless benefits are granted under the provisions of paragraphs 3, 4 or 5, at the time that the benefit
would be accorded. The term treaty benefits mainly consist of the treaty’s distributive and relief
rules, which limit the taxing rights of the contracting states (Articles 6–23 OECD-MC).34 In line with
US-style LoB clauses, the protection of residents under Article 24 OECD-MC is also considered a
30 See J. Bates et al., “Limitation on Benefits Articles in Income Tax Treaties: The Current State of Play” (2013), 41 Intertax 6/7, 395 - 407.
31 See S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse (Vienna: Linde, 2016) 355; L De Broe and J Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS) (2016), 227; D. Dominguez, “Limitation on Benefits: Comparison between the US LOB and the OECD LOB proposed under Action 6”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse (Vienna: Linde, 2016), 305.
32 See L De Broe and J Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146.
33 J.C. Fleming, Jr., “Searching for the Uncertain Rationale Underlying the US Treasury’s Anti-treaty Shopping Policy” (2012), 40 Intertax 4, 245-253; see also L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; B.M. Kerekes, “Limitation on Benefits Clauses – Function, Purpose and history” in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse (Vienna: Linde, 2016), 176.
The fact that 41 pages of the final report of Action 6 are devoted to the discussion of the proposed LoB clause is a perfect illustration of that complexity. In addition, the OECD itself admits in point 6 of Action 6 that “the administrative capacity of some countries might prevent them from applying certain detailed treaty rules and might require them to opt for more general anti-abuse provisions”.
34 See M. Lang, BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties (2014), 74 Tax Notes Int’l 7, 655-664; R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 226.
8
treaty benefit subject to the LoB clause.35 Not deemed a treaty benefit however is the protection
under Article 4(3) (corporate residence tie-breaker), Article 9(2) (downward transfer pricing
adjustment) and Article 25 OECD-MC (mutual agreement procedure).36 Furthermore, as a LoB clause
has a restrictive effect, it is only relevant if all other requirements of the treaty (e.g., being a resident,
being the beneficial owner, etc.) have been satisfied.37
Art X Paragraph 2 defines the qualified persons which are automatically entitled to all treaty benefits.
It has six subparagraphs, each of which contains a category of residents that are qualified persons by
reference to the attributes of these persons. In general, the concept of qualified persons comprises
a) individuals;
b) the two contracting states and subdivisions thereof;
c) entities that successfully meet the stock exchange test;
d) certain charitable organizations and pension funds;
e) entities that meet the ownership and base erosion test; and
f) certain collective investment vehicles.
The subparagraphs c) - f) are linked to specific tests, which have in common that, once they are
fulfilled, the person concerned is deemed to have a sufficient nexus with its residence state and the
treaty benefits fully apply.38 However, some of these subparagraphs (especially (c) and (e)) are
complex to administer and contain significant exceptions, which could impair the effectiveness of the
LoB clause in countering treaty shopping.39
The stock exchange test (Art X(2)(c)) is based on the assumption that because the shares of publicly-
traded companies are generally widely-held and are subject to stringent securities legislation, these
companies are unlikely to be specifically established for treaty shopping purposes.40 Additionally,
35 See F.A. Vega Borrego, Limitation on Benefits Clauses in Double Taxation Conventions, The Hague: Kluwer Law International, 2006, 95.
36 See L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 226.
37 See L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; L. De Broe, International Tax Planning and Prevention of Abuse, Amsterdam, IBFD, 2008, 739–741 and references there.
38 See R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 227; J. Bates et al., “Limitation on Benefits Articles in Income Tax Treaties: The Current State of Play”, 41 Intertax 6/7, 395 - 407; D. Dominguez, “Limitation on Benefits: Comparison between the US LOB and the OECD LOB proposed under Action 6”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 302 et seq.
39 See J.C. Fleming, Jr., “Searching for the Uncertain Rationale Underlying the US Treasury’s Anti-treaty Shopping Policy”, 40 Intertax 4, 245-253; R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 227.
40 See F.A. Vega Borrego, “Limitation on Benefits Clauses in Double Taxation Conventions”, The Hague, Kluwer Law International, 2006, 98; J. Bates et al., “Limitation on Benefits Articles in Income Tax Treaties: The Current State of Play”, 41 Intertax 6/7, 395 - 407; L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; D. Dominguez, “Limitation on Benefits: Comparison between the US LOB and the OECD LOB proposed under Action 6”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 305.
9
being listed on a stock exchange arguably requires a sufficient degree of nexus to a territory.
However, the stock exchange test contains many newly introduced terms needing legal definitions
(such as recognized stock exchange, principal class of shares, disproportionate class of shares). Some
of these terms are extensively defined in paragraph 6 of the LoB clause, some however will definitely
spark discussions, disagreement and dispute between taxpayers and tax administrations.41
The ownership and base erosion test (Article X(2)(e)) contains two cumulative subtests that both
need to be fulfilled. The two tests ensure that a majority of the equity (assessed by the ownership
test) owners and non-equity (evaluated by the base erosion test) holders are residents of one of the
contracting states. The ownership test prong of the proposed clause provides that at least 50% of
each class of shares must be owned, directly or indirectly, by qualified persons themselves. The base
erosion test prong is satisfied when less than 50% of the company’s gross income for the taxable
period is paid or accrued to non-qualified persons. The rationale of the base erosion test is that non-
equity holders (creditors) could just as much as equity owners influence the decision making of the
company and economically own a company.42 At a first glance, the mechanical nature of the
ownership and base erosion test seems to allow a quick and reliable detection of treaty abusive
structures; however it lacks clarity with respect to specific terms such as for example “gross
income”43 and thus will still provide opportunities for creative tax planners.44
Art X Paragraph 3 provides for an activity test, under which persons that would have been classified
as non-qualified persons according to the Paragraph 2 may still become entitled to the treaty
benefits. This however is only a limited entitlement to the treaty benefits only with respect to a
particular item of income and only if that item of income is derived in connection with the active
conduct of a business of that (otherwise non-qualified) person in its residence state, including
activities conducted by affiliated persons.45 The activity test is rather complex, as it requires each
time when income is obtained a verification of whether the recipient is engaged in the active conduct
41 See for a comprehensive list of terms and concepts in need of further interpretation R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 229 et seq; D. Dominguez, “Limitation on Benefits: Comparison between the US LOB and the OECD LOB proposed under Action 6”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 306 et seq.
42 See J. Bates et al., “Limitation on Benefits Articles in Income Tax Treaties: The Current State of Play”, 41 Intertax 6/7, 395 - 407; R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 236 et seq.
43 See further D. Dominguez, “Limitation on Benefits: Comparison between the US LOB and the OECD LOB proposed under Action 6”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 306 t seq; R. Szudoczky and P. Koch, “Limitation on Benefits: ‘Qualified Person’ – Article X (1) and (2) of the OECD Model”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 237 et seq.
44 See A. Wardzynski, ‘The Limitation on Benefits Article in the OECD Model: Closing Abusive (Undesired) Conduit Gateways’ (2014), 68 Bulletin for International Taxation 9, 478.
45 See further L.P. Lukjanenko, “LoB – Article X (3) of the OECD MC: Active Conduct of a Business”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse (Linde: Vienna, 2016), 210; R.J.S. Tavares, “The ‘Active Trade or Business’ Exception of the Limitation on Benefits Clause”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS) (Linde: Vienna, 2016), 139.
10
of a business in its residence state and the payment for which benefits are sought is related to that
business.46 Or, if the income is derived from a business activity conducted in the source state or from
a related person, the business activity in the residence state should be “substantial“47 compared to
the business activity generating the item of income in the source state.
The term “active trade or business” remains widely undefined in Art X and therefore must be
interpreted by reference to domestic law (Article 3(2) OECD-MC).48 However, the proposed LoB
clause limits the scope of the term “business” as it provides that the business of making or managing
investments for its own account cannot benefit from the activity clause, unless the relevant activities
are conducted by a bank, an insurance enterprise or a registered securities dealer.49
In summary, the active conduct of a business test implies economic substance in terms of premises,
personnel and activities.50 However, this does not exclude all forms of treaty shopping especially not
certain conduit arrangements.51 Therefore Action 6 additionally recommends that the LoB clause
should be supplemented by an anti-conduit rule, which would either be the Principal Purpose Test
(see below), or a separate treaty-based anti-conduit rule, or a domestic anti-abuse rule or judicial
doctrine that would achieve a similar result.
Art X Paragraph 4 is a so-called derivative benefits clause that allows certain entities owned by
residents of third states to obtain treaty benefits if these residents are “equivalent beneficiaries“ who
would have been entitled to equivalent benefits if they had invested directly in the source state. The
underlying rationale of the derivative benefits clause is that treaty shopping cannot arise when the
benefits provided by the specific tax treaty concerned would be available to the recipient in similar
ways had the income been directly remitted to the ultimate recipient.52 This exception to the base
46 See further L.P. Lukjanenko, “LoB – Article X (3) of the OECD MC: Active Conduct of a Business”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 212 et seq; R.J.S. Tavares, “The ‘Active Trade or Business’ Exception of the Limitation on Benefits Clause”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 142 et seq.
47 Action 6, Section A, pt. 54.
48 See further R.J.S. Tavares, “The ‘Active Trade or Business’ Exception of the Limitation on Benefits Clause”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 144 et seq; L.P. Lukjanenko, “LoB – Article X (3) of the OECD MC: Active Conduct of a Business”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 1 et seq; for interpretation of the term “business”, A. Rust, “Chapter 6 ‘Business’ and ‘Business Profits’”, in G. Maisto (Ed.), The Meaning of ‘Enterprise’, ‘Business’ and ‘Business Profits’ under Tax Treaties and EU Tax Law, (Amsterdam: IBFD, 2011), 85 et seq.
49 Action 6, Section A, pt. 48; see also R.J.S. Tavares, “The ‘Active Trade or Business’ Exception of the Limitation on Benefits Clause”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 145 et seq; L.P. Lukjanenko, “LoB – Article X (3) of the OECD MC: Active Conduct of a Business”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 215 et seq.
50 See also R.J.S. Tavares, “The ‘Active Trade or Business’ Exception of the Limitation on Benefits Clause”, in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 145 et seq; L.P. Lukjanenko, “LoB – Article X (3) of the OECD MC: Active Conduct of a Business”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 215 et seq.
51 See J.C. Fleming, Jr., “Searching for the Uncertain Rationale Underlying the US Treasury’s Anti-treaty Shopping Policy”, 40 Intertax 4, 245-253; A. Wardzynski, ‘The Limitation on Benefits Article in the OECD Model: Closing Abusive (Undesired) Conduit Gateways’, 68 Bulletin for International Taxation 9, 478.
52 See S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 357 et seq; M.A.C. Camayo, “Limitation on Benefits: Derivative Benefits and Discretionary Relief”, in D. Blum and M. Seiler (eds.),
11
erosion test is applicable in situations where the taxpayer who is a resident of state A receives
income from a company in state B which received the income from sources in state C, and the tax
treatment of the three-state structure is the same as it would have been had the taxpayer received
the income directly from state C.53 The derivate benefits clause contains again an ownership and
base erosion test. Whereas the base erosion test is the identical to the one of paragraph 2,54 the
ownership test is fulfilled if seven or fewer beneficiaries own (directly or indirectly) at least 95% of
the shares.55
Finally, Art X Paragraph 5 contains a so-called discretionary relief clause.56 Since the mechanical tests
of the previous paragraphs might neither be comprehensive nor perfect, the competent authority
may exercise discretion and grant benefits to taxpayers who fail these mechanical tests but
nevertheless should be afforded treaty benefits as the establishment, acquisition or maintenance of
a resident and the conduct of its operations did not have as one of their principal purposes the
obtaining of treaty benefits.57 Discretionary relief shall only be granted on request of the taxpayer.
The competent authorities may opt to accord all or a limited number of treaty benefits, and can only
deny relief after having received (non-binding) input from the competent authorities of the other
contracting state.58
2.2. Principal Purpose Test
Paragraph 7 of Article X contains a Principal Purpose Test, which, if the test is failed, denies certain
treaty benefits even if the LoB clause resulted in a full application of the treaty.59 This limitation of
the application of a tax treaty is derived from the treaty practice of the United Kingdom and several
Preventing Treaty Abuse (Vienna: Linde 2016), 233 et seq; A. Wardzynski, ‘The Limitation on Benefits Article in the OECD Model: Closing Abusive (Undesired) Conduit Gateways’, 68 Bulletin for International Taxation 9, 476.
53 See also M.A.C. Camayo, “Limitation on Benefits: Derivative Benefits and Discretionary Relief”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 234 et seq; S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 357 et seq.
54 For further details, see M.A.C. Camayo, “Limitation on Benefits: Derivative Benefits and Discretionary Relief”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 242 et seq.
55 Action 6, Section A, pt. 59 et seq; see further M.A.C. Camayo, “Limitation on Benefits: Derivative Benefits and Discretionary Relief”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 239 et seq.
56 See L. De Broe and J. Luts, "BEPS Action 6: Tax Treaty Abuse", 43 Intertax 2, 122-146; M.A.C. Camayo, “Limitation on Benefits: Derivative Benefits and Discretionary Relief”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 246 et seq.
57 See also J. Bates et al., “Limitation on Benefits Articles in Income Tax Treaties: The Current State of Play”, 41 Intertax 6/7, 395 - 407; M.A.C. Camayo, “Limitation on Benefits: Derivative Benefits and Discretionary Relief”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 249 et seq; S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 358 et seq.
58 Action 6, Section A, pt. 65.
59 See A. Bhargava, ”The Principal Purpose Test: Functioning, Elements and Legal Relevance” in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse (Vienna: Linde, 2016), 315 et seq.
12
East-Asian countries60, where so called “main purpose test” provisions have been included in
primarily the dividend, interest and royalty articles of many negotiated tax treaties in recent years.61
While the overall rationale of Article X(7) and of the main purpose test in the UK treaties might be
the same, there are some differences. Maybe only a semantic difference but the OECD proposes a
“principal purpose test” while the treaty practice contains “main purpose tests”. More significant is
Article X(7)’s intent to govern all benefits of a treaty and not only the withholding tax reduction on
passive income.62 The proposed Article X(7) reads as follows:
“Notwithstanding the other provisions of this Convention, a benefit under this Convention shall
not be granted in respect of an item of income or capital if it is reasonable to conclude, having
regard to all relevant facts and circumstances, that obtaining that benefit was one of the
principal purposes of any arrangement or transaction that resulted directly or indirectly in that
benefit, unless it is established that granting that benefit in these circumstances would be in
accordance with the object and purpose of the relevant provisions of this Convention.”
The proposed Principal Purpose Test actually contains two separate tests in order to determine
whether the benefit of the treaty should be granted in a specific case.63 The first test is a subjective
test: is obtaining the treaty benefit one of the principal purposes of any arrangement or transaction
that resulted directly or indirectly in that benefit? This subjective test is followed by a second,
objective, test: the treaty benefit can still be granted if granting that benefit would be in accordance
with the object and purpose of the relevant treaty provision(s).
While the proposed Principal Purpose Test provision has these two tests, it is obvious that the main
rule of Art X(7) is the subjective test. If the principal purpose of an arrangement or transaction is
obtaining the treaty benefit, the PPT applies. The second test (object and purpose of a treaty) is only
an exception to the main rule, which however could safeguard the application of the treaty. The
interplay of these two requirements or tests is not trivial. Both tests contain undefined terms and
expressions, and at first sight seem to use terms with different meanings synonymously.
The first criterion which is very difficult to apply in practice is the essential application requirement
for the proposed rule – the subjective criterion that “obtaining a benefit must be one of the principal
60 See below Section 4.
61 See P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416; M. Seiler, GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU (Vienna: Linde, 2016), 140 et seq; S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 359 et seq.
62 See P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416.
63 See further P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416; S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 360.
13
purposes.” The main problem of this criterion is the practical difficulties involved in proving an
intention.64 The final report of Action 6 though states that the question whether “obtaining a benefit
is one of the principal purposes” of a specific arrangement can be and needs to be answered by an
‘‘objective analysis of the aims and objects of all persons involved.’’65 The aim of this ‘‘objective
analysis’’ is to draw conclusions on the intention of the transacting party at issue. The result of the
Principal Purpose Test though depends exclusively on the intention and the proof of the intention of
the transacting parties. While such subjective criteria can always be deduced on the basis of external
facts, the underlying intentions and motives though are very difficult to prove.66 Therefore and with
good reasons, legislators abstain, when possible, from attaching fiscal consequences to the existence
of such an intention.
The main issue with such subjective rules67 is that the rules on the burden of proof effectively
determine the result of the test and not the actual intention.68 So if the tax authority has to prove
that one of the main objectives of the taxpayer was to obtain the treaty benefit, it is already fighting
a losing battle. Conversely, the taxpayer has no chance of fending off the accusation of abuse if it has
to present evidence that benefiting from one or several treaty provisions was not one of the primary
motives. Taxation and the application of a tax treaty would thus be determined solely by a
procedural question.
To, at least prevent bilateral disagreement on the outcome of the Principal Purpose Test due to
diverging domestic rules on the burden of proof, Article X(7) regulates it somewhat itself. The legal
consequences of Article X(7) apply if it is “reasonable to conclude” that one of the main objectives of
the taxpayer was to obtain the benefits of the tax treaty.69 At a first glance, this puts the burden of
proof into the tax authority’s hands, which has to draw this conclusion and justify it.70 However as it
must only be ‘‘reasonable’’ but not compelling, these requirements arguably are not overly
64 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; M. Seiler, GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU, 140 et seq.
65 Action 6 page 57.
66 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS) (Vienna: Linde, 2016), 284; S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 362; see further US Treasury Department, Technical Explanations to the US-Model Tax Treaty 2006, Art 22.
67 One current example in treaty law is Article 19(1)(b)(ii) OECD-MC, according to which the state of residence has the right to tax income from government when the services are rendered in the residence state and the individual “did not become a resident of that State solely for the purpose of rendering the services.”
68 See E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), Base Erosion and Profit Shifting (BEPS), 284; M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 208.
69 See P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416.
70 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 284 et seq.
14
demanding. Therefore, the tax authority does not need to produce full evidence.71 While Article X(7)
thus attempts to establish a balance between the interests of the tax authority and those of the
taxpayer, the bias in favor of the tax authority is however fairly obvious.72 In practice, presenting
unequivocal evidence of the motives will therefore not be necessary and tax authorities will be
tempted to presume intention simply because of the presence of the benefit.
The final report of Action 6 makes it rather straightforward for tax administrations to assume an
abuse. It is by no means required that the sole purpose of the arrangement/transaction must aim at
obtaining a tax benefit. It does not even have to be the essential, the principal, or the main
purpose.73 Instead, it suffices if one of the main purposes of an arrangement/transaction is to obtain
a benefit. Thus, the rule assumes that not merely one main purpose but two or even multiple main
purposes may exist for a specific transaction, arrangement or structure. So, even if the taxpayer can
prove that the arrangement chosen is (also) motivated by non-fiscal reasons, the tax authority can
rebut that and the anti-abuse rule would apply.74 The rule would even apply if the taxpayer’s main
purpose was a non-fiscal one. It remains unclear how tax administrations and eventually courts
would distinguish between main purposes and secondary purposes, but also between different main
purposes.75 This distinction, however, is critical for the application of Article X(7) and thus for the
foreseen legal consequences. This is even more critical as tax treaties are interpreted by different
national courts. National courts may reach completely different judgments (with respect to the same
treaty and even with respect to the same case) and they are often unable to free themselves from
the fiscal interests of their state.76 Arbitrary decisions by tax authorities, an increase in the number
of court proceedings, diverging tax treaty interpretations by these courts and a reduction of legal
certainty might be the result. The result would be increased instances of double taxation and a
71 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 284; P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416.
72 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416; S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 363.
73 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 286; P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416; M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 209; A. Bhargava, ”The Principal Purpose Test: Functioning, Elements and Legal Relevance” in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 319.
74 See M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 209; M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664.
75 See E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 287; M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; A. Bhargava, ”The Principal Purpose Test: Functioning, Elements and Legal Relevance” in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 319 et seq.
76 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664.
15
further erosion of the desired uniformity in application of tax treaty provisions – the actual opposite
of what was the aim of the OECD BEPS Action Plan.
While it is highly unclear how to define and evidence that “one of the principal purposes” was to
obtain a treaty benefit, it is also unclear what treaty benefit is in question. At first, it seems fairly
obvious that the “treaty benefit” must be a benefit resulting from the application of the treaty itself
and not some benefit granted on the basis of domestic law or a different treaty.77 All treaty
provisions that do not effectively change the taxpayer’s tax position, such as the definitions in Article
3 or Article 4 of the OECD-MC do not by themselves lead to any benefits for the taxpayer.78 Even the
rules on the personal and substantive scope of application of the tax treaty (Articles 1 and 2 OECD-
MC) do not convey a benefit.79 In the end, such a benefit must result either from the distribution
rules (Articles 6 to 8 and 10 to 21) and usually in the source state, or from the methods article (Article
23A/B OECD-MC) in the residence state.80
With a “benefit of the treaty”, Action 6 obviously means that the tax position would have to improve
for the taxpayer as a result of the application of one or several treaty provisions as compared with
the domestic law.81 But does this refer to the overall tax burden in both contracting states combined
or does it refer to the difference between tax burdens in one contracting state with or without
application of the treaty?82 For example, if the structure chosen by the taxpayer aims at a reduction
of withholding taxes, this would reduce the tax burden for the taxpayer in the source state but not
necessarily the overall tax burden. The reduced withholding tax only has an impact on the
distribution of the taxation rights between the two states.83 The taxpayer benefits only marginally
77 See E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 278 et seq; A. Bhargava, ”The Principal Purpose Test: Functioning, Elements and Legal Relevance” in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 316 et seq.
78 See E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 279.
79 Also the protection from discrimination (Art 24 OECD-MC) or the authorization to initiate a mutual agreement procedure and subsequently an arbitration procedure (Art 25 OECD-MC) do not contain a “benefit of the treaty”. The application of the rules on information exchange and mutual enforcement assistance (Art 27 OECD-MC) can also not be seen as a “benefit of the treaty” even if some states (especially EU-Member States) attach certain benefits to the condition that comprehensive mutual administrative and enforcement assistance applies to a specific state in their domestic tax laws. In this case, however, a benefit resulting from domestic law is under consideration and not a treaty benefit, so that for this reason the proposed rule of Article X(7) cannot be effective; see E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 279; M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664.
80 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; A. Bhargava, ”The Principal Purpose Test: Functioning, Elements and Legal Relevance” in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 316 et seq; E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 279 et seq.
81 See also M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 214.
82 See E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 280; A. Bhargava, ”The Principal Purpose Test: Functioning, Elements and Legal Relevance” in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 317.
83 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664.
16
(i.e., the reduced source country withholding tax) while the residence state obtains a benefit in the
form of a lower foreign tax credit allowance to the taxpayer and therefore a higher tax revenue from
the source country payment. Whether it is sufficient to assume a “benefit” for the taxpayer in a
scenario where the structure only reduces compliance cost and possibly financing costs such as in the
example above is debatable.
Summing up, the Principle Purpose Test proposed in Article X(7) in Action 6 is a highly complex and
very vaguely written rule. This type of subjective test focusing on the intentions of the taxpayer has
been broadly criticized in the literature.84 It contains a range of terms and concepts that need further
interpretation and clarification. It is neither clear what determines a “principal purpose” nor what
could be considered a “treaty benefit”. Furthermore, the rule potentially shifts the burden of proof
towards the taxpayer who then has to evidence that the actions taken were not aimed at obtaining a
certain treaty benefit. It is rather trivial that furnishing proof that an intention did not exist when
such an intention might be deducible from the outcome of the transaction is virtually impossible. The
tax administration could easily assume that it is “reasonable to conclude” that one of the principal
objectives of the taxpayer was to obtain the benefits of the tax treaty. For taxpayers, and also for tax
administrations, this inevitably will lead to more legal uncertainty and higher compliance and
litigation costs.
3. Anti Treaty Abuse measures – historical and current perspectives
3.1. Evolution of the U.S. LoB clause
The United States first started in 1939 to conclude tax treaties with important trading partners
aiming to encourage international investment flows by reducing the burden of double taxation.85
While the first tax treaties such as, for instance, the 1948 U.S.-Netherlands Treaty86 applied only to
the treaty partner itself and not to their overseas possessions,87 during the 1950s most of these
84 See M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; P. Baker, “The BEPS Action Plan in the Light of EU Law: Treaty Abuse”, 60 British Tax Review 3, 408-416; M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 312; A. Bhargava, ”The Principal Purpose Test: Functioning, Elements and Legal Relevance” in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 324 et seq; E. Pinetz, “Use of a Principal Purpose Test to Prevent Treaty Abuse” in M. Lang et al (eds.), “Base Erosion and Profit Shifting (BEPS)”, 296 et seq; S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), “Preventing Treaty Abuse”, 361 et seq.
85 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325; M.F. Huber and M.S. Blum, “Limitation on Benefits Under Article 22 of the Switzerland-U.S. Tax Treaty”, 39 Tax Notes Int’l 6, 547-568; S.I. Roberts, “Legislative History Of United States Tax Conventions”, 1961.
86 U.S.-Netherlands tax treaty, signed on April 29, 1948.
87 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325.
17
treaties with European trading partners were extended to their overseas possessions. After their
independence many of those former European colonies, especially the former British colonies of the
Caribbean had become tax havens.88 Thus, by rewriting their domestic tax laws, some countries took
advantage of the extension of the U.S. tax treaties,89 which was the beginning of what is now
commonly referred to as “treaty shopping”90. From that point on, U.S. lawmakers became concerned
about that new phenomenon, stating that an income tax treaty providing benefits that can be
claimed by all residents of the other contracting state would pose a high risk of abuse by third-
country residents91 interposing a conduit company to invest in the U.S. and thus be indirectly eligible
to the treaty benefits, leading to the effect that other countries would have reduced incentives to
enter into treaty negotiations with the United States.92
The first foundations of what would later become the U.S. standard Limitation on Benefits provision,
which is nowadays included in almost all U.S. tax treaties, however dates back to the 1945 U.S.-U.K.
tax treaty.93 That treaty provided for a general (15%)94 and a special (5%)95 withholding tax rate on
dividends. In order to qualify for the special 5% withholding tax rate, both an ownership test and an
active business test had to be met.96 However, even if these two tests were met, the reduced
withholding tax rate of 5% would not have been applicable in case “the relationship of the two
corporations has been arranged or has been maintained primarily with the intention of securing such
reduced rate.”97 This so called “arranged or maintained” test aimed at ensuring that only legitimate
corporations benefitted from the reduced withholding tax rate.98 The rule was purely subjective as it
was based solely on the taxpayer’s intent and therefore difficult to enforce.99 The “arranged or
88 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325.
89 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325.
90 See M.F. Huber and M.S. Blum, “Limitation on Benefits Under Article 22 of the Switzerland-U.S. Tax Treaty”, 39 Tax Notes Int’l 6, 547-568; R.S. Avi-Yonah and H.J.I. Panayi, “Rethinking Treaty Shopping: Lessons for the European Union”, University of Michigan Law School Working Paper No. 182 (2010).
91 Third-country residents whose country of residence has no tax treaty with the U.S. or do have one but with less beneficial rates.
92 M.F. Huber and M.S. Blum, “Limitation on Benefits Under Article 22 of the Switzerland-U.S. Tax Treaty”, 39 Tax Notes Int’l 6, 547-568; J.C. Fleming, Jr., “Searching for the Uncertain Rationale Underlying the US Treasury’s Anti-treaty Shopping Policy”, 40 Intertax 4, 245-253.
93 Convention Between the United States of America and the United Kingdom of Great Britain and Northern Ireland for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, signed on Apr. 16, 1945 (1945), Art. VI. See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping,” 38 Tax Executives 325; I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers,” 17 Fordham Int'l L.J. 3, 776-824.
94 1945 U.S. – U.K. tax treaty, Art. VI(1).
95 1945 U.S. – U.K. tax treaty, Art. VI(1).
96 1945 U.S. – U.K. tax treaty, Art. VI(1).
97 1945 U.S. – U.K. tax treaty, Art. VI(1).
98 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers,” 17 Fordham Int'l L.J. 3, 776-824.
99 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325 (1986).
18
maintained” rule though appeared in several subsequent U.S. income tax treaties signed in the
1950s.100 The 1945 U.S.-U.K. treaty thus can be seen as the first precursor to Article X(7) of Action 6.
The 1962 U.S.-Luxembourg treaty101 contained the first separate anti-treaty shopping provision
specifically aimed at limiting benefits under the treaty to those persons who were citizens or residents
of one of the contracting states102 by disallowing any treaty benefits to “any holding company entitled
to any special tax benefit under Luxembourg Law (…), or to any income derived from such companies
by any shareholder thereof.”103 Thus, this approach provided a renunciation from the purely subjective
“arranged or maintained” test towards a more objective test, which disregards the motives for
establishing the structure.104 However, due to the absence of such a provision in other treaties of this
time, Article 15 of the 1962 U.S.-Luxembourg treaty was not successful in preventing treaty shopping
overall as taxpayers were using alternative holding locations.105 While the U.S. negotiated new tax
treaties with a number of countries over the following years,106 none of these treaties contained a
limitation on benefits clause of any sort,107 with the exceptions of the Treaty with Trinidad and
Tobago108 and with Finland109 in 1970.
In 1977, the United States published a Model Treaty110 to act as a coherent guide for future treaty
negotiations.111 The 1977 U.S. Model Treaty included a Limitation on Benefits article, which denied
100 E.g., Austria, Denmark, Switzerland, Ireland.
101 Convention Between the United States of America and the Grand Duchy of Luxembourg with Respect to Taxes on Income and Property, signed on Dec. 18, 1962 (1962 U.S.-Luxembourg tax treaty).
102 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers”, 17 Fordham Int'l L.J. 3, 776-824; Grady, “Income Tax Treaty Shopping: An Overview of Prevention Techniques”, (1983) 5 Northwestern Journal of International Law & Business 3, 626-657.
103 1962 U.S.-Luxembourg tax treaty, Art. XV.
104 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); Grady, “Income Tax Treaty Shopping: An Overview of Prevention Techniques”, 5 Northwestern Journal of International Law & Business 3, 626-657.
105 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers”, 17 Fordham Int'l L.J. 3, 776-824; Grady, “Income Tax Treaty Shopping: An Overview of Prevention Techniques”, 5 Northwestern Journal of International Law & Business 3, 626-657.
106 Japan (1962), Sweden (1963), Belgium (1965), Germany (1965), the Netherlands (1965), the United Kingdom (1966), Canada (1966), Trinidad and Tobago (1966), and France (1969).
107 Notably, the U.S. government had not completely given up developing Limitation on Benefits provisions during the 1960s. Two proposed treaties (Israel, 1965 and Brazil, 1967), none of which entered into force, included separate Limitation on Benefits provisions that were similar to Article 15 of the 1962 U.S.-Luxembourg treaty, but made various refinements (see further H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936)).
108 The Convention between the Government of the United States of America and Trinidad and Tobago for the Avoidance of Double Taxation, the Prevention of Fiscal Evasion with respect to Taxes on Income, and the Encouragement of International Trade and Investment, Jan. 9, 1970, 22 U.S.T. 164 (hereinafter “1970 U.S.-Trinidad and Tobago Treaty”).
109 The Convention between the Government of the United States of America and the Government of the Republic of Finland for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with respect to Taxes on Income and on Capital, Mar. 6, 1970,, 22 U.S.T. 40, T.I.A.S. No. 7042 (“1970 U.S.-Finland Treaty”).
110 U.S. Model Income Tax Convention of 1977 (“1977 U.S. Model Treaty”).
111 See I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers”, 17 Fordham Int'l L.J. 3, 776-824; see 1977 U.S. Model Treaty.
19
treaty benefits to a company resident in a contracting state if more than 25% of the company’s capital
is owned by non-residents and if by reason of special measures the dividend, interest, or royalty
income of the company is taxed at a substantially lower rate than its regular corporate profits.112 This
approach of the two-part “special measures” and “foreign ownership” test was however only effective
when a corporation was subject to a substantially reduced special tax rate compared to the generally
applicable tax rate.113 Thus, in case a country’s tax system imposed very low corporate tax rates in
general, this Limitation on Benefits article did not provide for effective countering of treaty abuse.114
Until the publication of the 1996 U.S. Model Income Tax Convention, several specific bilateral treaties
as well as the 1981 U.S. Model Treaty brought some important developments. The 1978 Protocol115
to the 1968 U.S.-France treaty116 introduced a so called derivative benefits rule. Under this rule,
treaty benefits were still granted to nonresident shareholders where comparable benefits were
available to them under another treaty with the source country.117 Further, the first “public
company” exception to the ownership test as well as the first anti-conduit limitation to treaty
benefits were included in that 1978 Protocol.118 The 1980 U.S.-Jamaica tax treaty119 was the first
treaty to include a base erosion test, stating that a company’s income could not be “used in
substantial part to meet liabilities to persons who are residents of a State other than a Contracting
State”.120 Furthermore, Article 17 of the 1980 U.S.-Jamaica tax treaty was the first Article expressly
entitled “Limitation on Benefits”. Notably, this treaty was also the first treaty in force to include a
generally applicable “public company” test.121
112 See M.F. Huber and M.S. Blum, “Limitation on Benefits Under Article 22 of the Switzerland-U.S. Tax Treaty”, 39 Tax Notes Int’l 6, 547-568.
113 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936).
114 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers,” 17 Fordham Int'l L.J. 3, 776-824.
115 1978 Protocol to the U.S.-France Income Tax Convention, Art. 1(1).
116 Convention Between the United States of America and the French Republic with Respect to Taxes on Income and Property, signed on Jul. 28, 1967 (1967 U.S.-France tax treaty).
117 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936).
118 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936).
119 Convention Between the Government of the United States of America and the Government of Jamaica for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, signed on May 21, 1980 (1980 U.S.-Jamaica tax treaty).
120 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936).
121 The 1978 U.S. – France Protocol contained such a test with regard to corporations receiving income from shipping and air transport only.
20
The 1981 U.S. Model Treaty122 significantly changed the Limitation on Benefits clause.123 The scope
was extended to trusts and other entities.124 The denial of treaty benefits was extended to all forms
of income, not only passive income.125 The limitation on benefits was based on foreign ownership
(less than 25% foreign ownership) and special measures.126 A base erosion test was added which
denied treaty benefits if a substantial part of the income was paid to residents of a third country as
interests, royalties or other deductible payments.127 Additionally, two safe harbor provisions were
added – one for publicly traded companies128 and one for taxpayers who could prove that the
principal purpose for establishing the company structure was not aimed at receiving the treaty
benefits.129 Since the release of the 1981 U.S. Model Treaty, a LoB article has been included in every
U.S. tax treaty entered into after the release.
Besides changes in the details and structure of the wording of the respective limitation on benefits
clauses found in U.S. treaties, the 1982 U.S.-New Zealand130 and the 1982 U.S.-Australia131 tax treaties
contained a new provision specifically related to income derived by a trustee. With the U.S.-Italy Tax
Treaty Protocol in 1984, the U.S. tax treaty policy with regards to the ownership threshold (75%) was
liberalized as this high ownership requirement was seen as an obstacle to bona fide structures forming
a real business.132 Thus a 50% ownership threshold was introduced and included in tax treaties
negotiated and signed throughout the 1990s.133
122 United States Dept. of Treasury Model Income Tax Treaty, June 16, 1981.
123 See M.F. Huber and M.S. Blum, “Limitation on Benefits Under Article 22 of the Switzerland-U.S. Tax Treaty”, 39 Tax Notes International 547 (2005); I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers,” 17 Fordham Int'l L.J. 3, 776-8247.
124 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325 (1986).
125 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325 (1986).
126 See term “and” following paragraphs 1(a) and 1(b); the term “special measures” has been removed from the text.
127 See R.S. Avi-Yonah and O. Halabi, “US Treaty Anti-Avoidance Rules: An Overview and Assessment“, 66 Bulletin For International Taxation 4/5, 236-242.
128 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325 (1986); I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers”, 17 Fordham Int'l L.J. 3, 776-824.
129 See R.J. Rolfe and T.S. Doupnik, “The United States Attempts to Crack Down on Treaty Shopping”, 38 Tax Executives 325.
130 Convention between the United States of America and New Zealand for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with respect to Taxes on Income, together with a Related Protocol, signed on Jul. 23, 1982 (“1982 U.S.-New Zealand tax treaty”), Art. 16.
131 Convention Between the Government of the United States of America and the Government of Australia for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, signed on Aug. 6, 1982 (”1982 U.S.-Australia tax treaty”), Art. 16.
132 See R.S. Avi-Yonah and O. Halabi, “US Treaty Anti-Avoidance Rules: An Overview and Assessment“, 66 Bulletin For International Taxation 4/5, 236-242.
133 See R.S. Avi-Yonah and O. Halabi, “US Treaty Anti-Avoidance Rules: An Overview and Assessment“, 66 Bulletin For International Taxation 4/5, 236-242.
21
The Limitation on Benefits article in the 1989 U.S.-Germany tax treaty134 was the first article
representing all elements of a modern limitation on benefits clause and was received by the tax
community as a major innovation135 and has been used as a model for subsequent treaty
negotiations.136 Under that provision, three alternative methods for qualifying for treaty benefits were
provided. First, under the so called “automatic qualification”, five classes of persons, including
individuals, the U.S. or German government, public companies, non-profit organizations137 and persons
satisfying a more than 50% ownership test as well as a base erosion test, were automatically eligible
to treaty benefits.138 The second alternative was the so called “Active Business Connection Test” on a
subjective basis which replaced the “Principal Purpose Test” of prior treaties.139 In order to be entitled
to treaty benefits under the Active Business Connection Test a person must be engaged in the active
conduct of a trade or business in the residence state and, additionally, the income derived from the
other State must be connected with that trade or business.140 As a third alternative, the provision
contains a “safety-valve” under which the sovereign parties to the treaty are allowed to grant treaty
benefits on a discretionary basis, after considering all relevant facts and circumstances of the case.141
The 1996 U.S. Model Treaty included an amended Limitation on Benefits clause (Article 22).142 It
repealed the Principal Purpose Test, which was based on the taxpayer’s intent, which a tax
administration is usually ill equipped to identify, altogether.143 In order to avoid such a difficult burden
of proof for the tax administration, the amended Article 22 contained a series of objective tests.144 The
134 Convention Between the United States of America and the Federal Republic of Germany for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and Capital amd to Certain Other Taxes, together with a Related Protocol, signed on Aug. 29, 1989 (“1989 U.S.-Germany tax treaty”).
135 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); D.M. Berman and J.L. Hynes, “Limitation on Benefits Clauses in U.S. Income Tax Treaties”, 29 Tax Mgmt. Int'l J. 12, 692-721.
136 The United States concluded new tax treaties with Finland (signed on Sep. 21, 1989), Spain (signed on Feb. 22, 1990), Russia (signed on Jun. 17, 1992), Czech Republic (signed on Sep. 16, 1993), Slovak Republic (signed on Oct. 8, 1993), Kazakhstan (signed on Oct. 19, 1993), Ukraine (signed on Mar. 4, 1994), Sweden (signed on Sep. 1, 1994) and Portugal (signed on Sep. 6, 1994) in the years between 1989 and 1994. With minor exceptions, the Limitation on Benefits provisions included in these treaties mirror the one included in the 1989 U.S.-Germany tax treaty. The 1992 U.S.-Netherlands tax treaty and subsequently the 1994 U.S.-France tax treaty (signed on Aug. 31, 1994) though contain a much more detailed but also more complex wording, which was first insisted on by the Netherlands who were concerned that the text of the 1989 U.S.-Germany tax treaty was too general. See I.K. Sugarman, “The U.S.-Netherlands Income Tax Treaty: Closing the Doors on the Treaty Shoppers”, 17 Fordham Int'l L.J. 3, 776-824. For further discussion of the limitation on benefits provision in the 1992 U.S.-Netherlands tax treaty, see P.T. Kaplan, “Treaty Shopping Under the New U.S.-Netherlands Treaty” (1993), 47 Bulletin for International Fiscal Documentation 4, 175-180.
137 The inclusion of non-profit organizations was new to U.S. tax treaties.
138 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936).
139 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936).
140 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936).
141 See H.J. Levine and M.J. Miller, “U.S. Income Tax Treaties – The Limitation on Benefits Article” (Portfolio 936); Memorandum of Understanding to the 1989 U.S.-Germany tax treaty (8/29/89), Ex. VII.
142 The tax treaties concluded with Estonia (signed on Jan. 15, 1998), Latvia (signed on Jan. 15, 1998), Lithuania (signed on Jan. 15, 1998) as well as with Slovenia (signed on June 22, 2001) and Italy (signed on 25 Aug, 1999) included Limitation on Benefits provisions, all of which were following, for the most part, the 1996 U.S. Model Treaty.
143 Technical Explanations to the 1996 U.S. Model Treaty.
144 Technical Explanations to the 1996 U.S. Model Treaty.
22
rationale underlying each of the tests is that a taxpayer who is able to satisfy one or more of these
requirements probably has a real business purpose in respect of the established structure or has a
sufficiently strong nexus to the other contracting state that outweighs any purpose to obtain the treaty
benefits.145
Article 22 of the 2006 U.S. Model Treaty included significant changes to the former Limitation on
Benefits clause under the 1996 U.S. Model treaty, intending to make it more difficult for third-country
residents to benefit inappropriately from tax treaties between two countries.146 Article 22 of the 2006
U.S. Model Treaty again, just like the approach adopted by the 1996 U.S. Model Treaty, contained a
list of persons that were entitled to treaty benefits with no restrictions (individuals, publicly traded
companies, governments, political subdivisions and local authorities of a contracting state) – however
the scope of that list is much narrower as entities owned by the treaty state and governmental pension
funds are excluded.147 Further, the safe harbor rule of the 1996 U.S. Model Treaty for a substantial
trade or business has been removed, leaving substantiality to be determined solely on a facts and
circumstances basis.148 And, the 2006 U.S. Model Treaty does not include special derivative benefits
rules or rules for triangular arrangements anymore. However, the subsequently agreed treaties and
protocols with Bulgaria,149 Malta,150 New Zealand151 and France152 all provide for a derivative benefits
rule.
On 17 February 2016, the US Treasury Department released a revised US Model Income Tax
Convention (“2016 U.S. Model Treaty”).153 The 2016 U.S. Model Treaty includes a further improved
LoB clause (Article 22), which contains some of the developments in treaty practice from the
145 Technical Explanations to the 1996 U.S. Model Treaty; L. Freitas de Moreas e Castro, “US Policy to Counter Treaty Shopping – From Aiken Industries to the Anti-Conduit Regulations: A Critical View of the Current Double-Step Approach from the Perspective of Treaty Objectives and Purposes” (2012), 66 Bulletin for International Taxation 6, 2012, 300-312.
146 R.S. Avi-Yonah and O. Halabi, “US Treaty Anti-Avoidance Rules: An Overview and Assessment“, 66 Bulletin For International Taxation 4/5, 236-242.
147 R.S. Avi-Yonah and O. Halabi, “US Treaty Anti-Avoidance Rules: An Overview and Assessment“, 66 Bulletin For International Taxation 4/5, 236-242.
148 R.S. Avi-Yonah and M.B. Tittle, “The United States Model Income Tax Convention” (2007), 61 Bulletin for International Taxation 6, 224-234.
149 Convention between the Government of the United States of America and the Government of the Republic of Bulgaria for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with respect to Taxes on Income (signed on Feb. 23, 2007).
150 Convention between the Government of the United States of America and the Government of Malta for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with respect to Taxes on Income (signed on Aug. 8, 2008).
151 Protocol between the United States of America and New Zealand signed at Washington on December 1, 2008 Amending the Convention and Protocol between the United States of America and New Zealand for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with respect to Taxes on Income signed at Wellington on July 23, 1982.
152 Protocol Amending the Convention between the Government of The United States of America and the Government of The French Republic for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and Capital, Signed at Paris on August 31, 1994, as Amended by the Protocol Signed on December 8, 2004 (signed on Jan. 13, 2009).
153 Available for download at https://www.treasury.gov/resource-center/tax-policy/treaties/Documents/Treaty-US%20Model-2016.pdf (Feb. 25, 2017).
23
previous 10 years. Article 22(2)(f) of the 2016 U.S. Model Treaty reintroduces the derivative benefits
test which, if the more stringent conditions (relative to the past) are satisfied, maintains treaty
benefits for a company which is held at least 95% by (seven or fewer) non-residents of either
contracting state as long as this third state also has a comprehensive tax treaty with the United
States.154 The derivative benefits allowance however is also subject to the base erosion test, which
denies the benefits if 50% or more of gross income of this company is paid or accrued in form of
deductible payments to “bad recipients” (e.g., non treaty residents). The 2016 U.S. Model Treaty
adds to the LOB article a provision that permits companies that serve as the active headquarters
company of a multinational corporate group (“headquarters companies”) to receive treaty benefits
though limited to withholding tax reductions on dividends and interests. This new headquarters
companies test (Article 22(5)) requires a holding company to exercise primary management and
control functions (and not just supervision / administration) in its residence state with respect to
itself and its geographically diverse subsidiaries.155 Additionally, the headquarters test also includes a
base erosion test. For publicly traded companies the fulfillment of the stock exchange clause gets
more difficult as Article 22(2)(c)(i) requires that the stock exchange must be in the same country as
the company is in. This rule is intended to address “inversion” transactions in which U.S. companies
inverted to tax favorable jurisdictions such as for example Bermuda, held the board meetings in
Barbados to qualify under the US-Barbados treaty, and claimed exemption from the LoB clause
because they were publicly traded on the New York Stock Exchange.156 Further, the 2016 U.S. Model
Treaty makes some changes in the detailed wording of the active trade or business test (Article
22(3)).
3.2. The U.K. Main Purpose Test
The concept of a “main purpose test” as a measure against treaty shopping has long been present in
U.K.’s domestic and international tax law and still it is very scarcely researched in the international
literature. Since as early as the 1920s, the United Kingdom has had anti-avoidance legislation that
targeted certain types of transactions, i.e., specific anti-avoidance provisions. The distinctive feature
154 Previous tax treaties (e.g., U.S.-Germany of 1989, U.S.-UK of 2001, U.S.-Poland of 2013) contained similar provisions, which however limited the scope of admissible third states to EU/EAA Member States or parties to the NAFTA agreement.
155 See also Article 16(1)(h) of the US-Austria income tax treaty of 1998.
156 See R.S. Avi-Yonah, “Full Circle? The Single Tax Principle, BEPS, and the New US Model”, University of Michigan Law & Economics Research Paper Series No 15-019.
24
of these provisions was that it generally tested the motive of the taxpayer behind the transaction.157
Further on, in the 1960s the domestic income tax law provided for a rule to disregard transactions
that did have “as their main object, or one of their main objects, to enable tax advantages to be
obtained”158. The rule was an anti-avoidance rule aimed at dividend-stripping and bond-washing
transactions, which had proved resistant to specific anti-avoidance rules. To not impact legitimate
commercial transactions, the legislation would not affect bona fide commercial transactions or those
carried out in the normal course of making or managing investments, unless the (or one of the) main
object(s) of the transactions was tax avoidance.159 The main purpose test next appeared in 1977 – to
counter capital gains tax avoidance schemes,160 it denied tax relief if a reorganization or
reconstruction was “part of a scheme or arrangements of which the main purpose, or one of the main
purposes, is avoidance of liability to capital gains tax or corporation tax“. Again, a bona fide
commercial reason was required for the transactions. Nowadays, similarly worded specific anti-
avoidance provisions can be found throughout the UK domestic income tax system.161
As with the proposed “Principal Purpose Test“ of Action 6, the key issues with the “main purpose
test“ evolve around the question of how to identify the purpose of a particular arrangement and
what purpose is the “main purpose“. In this respect, the British case law has developed some
guidance. First, the “main purpose test“ does not automatically assume that every transaction is
exclusively performed for the purpose of tax avoidance.162 Secondly, the mere existence of a tax
advantage and also the knowledge that a certain structure will lead to a tax advantage (even if the
structure has been recommended by tax advisors) also does not automatically constitute a “main
purpose“.163 However, the relative size of the tax advantage in relation to the transaction volume
seems to be one of the important factors. If the tax advantage is a mere "icing on the cake" it will not
constitute a main purpose.164 The “main purpose test“ is thus a question of whether the tax position
was an end in itself, or just an incidental improvement to the bona fide commercial transaction.165
157 See D. Roxburgh, “UK Anti-Avoidance – Where Are We Now?” (2015), 55 European Taxation 2/3, 88-95.
158 Sec. 28 FA 1960.
159 See R. Shiers and P. Miller, “The Whole or Main Purpose Test”, 2007 Tax Journal 876, 8-12.
160 See the court rulings in Floor v Davis 52 TC 609; Furniss v Dawson 55 TC 324; Craven v White 62 TC 1.
161 See R. Shiers and P. Miller, “The Whole or Main Purpose Test”, 2007 Tax Journal 876, 8-12.
162 See Brebner v IRC 43 TC 705: “it would be quite wrong as a necessary consequence to draw the inference that, in adopting [a structure which involves a reduced tax liability] one of the main objects is, for the purposes of the section, avoidance of tax.”
163 See Lewis (Trustee of the Redrow Pension Scheme) v IRC [1999] STC (SCD) 349.
164 See Trustees of the Sema Group Pension Scheme v IRC [2003] STC 95.
165 This is also confirmed in Snell v HMRC [2006] All ER (D) 336, which looked at the structuring of the disposal of a company in temporal conjunction with an emigration. The structure allowed for the avoidance of U.K. capital gains taxes. While the purpose for the disposal of the company itself was the plan to emigrate, the main purpose for the structuring of the transfer was the tax saving. Mr. Snell “had the purpose of becoming non-resident before redeeming the loan notes and accordingly that one of his main purposes (indeed the only main purpose) of effecting the arrangements was avoidance of capital gains tax”.
25
At the tax treaty level, already in 1945 the U.S.-U.K. tax treaty used the term “primarily with the
intention” in order to determine which arrangements of the taxpayer should not be allowed
application of the reduced tax treaty rates.166 In the 1970s the United Kingdom started to generally
include this specific anti-treaty shopping rule into their double tax treaties. The first of such
provisions from this era could be found in the 1976 U.K.-Ireland Tax Treaty.167 These provisions are
usually not separate treaty articles but are included in specific distribution rules, mainly into the
passive income articles (dividend, Interest, royalties) and in the Other Income article. According to
these anti-abuse rules, the specific article(s) of the treaty shall not apply if “it was the main purpose
of a person concerned with the creation or assignment of the shares [or debt claims, or similar assets
giving rise to income] in respect of which the dividend [or interest, etc.] is paid to take advantage of
this Article by means of that creation or assignment”.
In 2013, the United Kingdom introduced a domestic General Anti-Avoidance Rule (GAAR) which
further refines the main purpose test.168 The GAAR operates independently of existing anti-avoidance
provisions and in priority to any other anti-avoidance rules. This domestic GAAR prohibits “abusive
tax arrangements” that are determined by a series of intertwined subjective tests. Section 207 FA
2013 defines arrangements as abusive if
(1) “it would be reasonable to conclude that the obtaining of a tax advantage was the main
purpose, or one of the main purposes, of the arrangements” and
(2) “the entering into or carrying out of [the arrangement] cannot reasonably be regarded as a
reasonable course of action in relation to the relevant tax provisions […]”.
Sec 207(4) FA further provides examples “of something which might indicate that tax arrangements
are abusive”:
a) the arrangements result in an amount of income, profits or gains for tax purposes that is
significantly less than the amount for economic purposes;
b) the arrangements result in deductions or losses of an amount for tax purposes that is
significantly greater than the amount for economic purposes; and
c) the arrangements result in a claim for the repayment or crediting of tax (including foreign
tax) that has not been, and is unlikely to be, paid.
166 See S. Kolundzija, “OECD Minimum Standard: Comparing LOB and PPT”, in D. Blum and M. Seiler (eds.), Preventing Treaty Abuse, 359.5.
167 See W. Cui, “The China-United Kingdom Income Tax Treaty (2011)” (2013), 67 Bulletin for International Taxation 6, 271-279.
168 See further D. Roxburgh, “General Anti-Abuse Rule (GAAR)” (2014), 54 European Taxation 2/3, 113-116.
26
With no cases decided yet on grounds of the GAAR, it is impossible to estimate how courts will deal
with the rule and how the preceding case law will influence the interpretation of the “main purpose
test” in the 2013 U.K. GAAR. The GAAR though is accompanied with a large body of explanatory
materials. The so-called GAAR Guidance drafted and published by HMRC is thereby an important
source for understanding the UK GAAR.169 The Guidance is important as it does not merely reflect
HMRC’s opinion on the GAAR. Rather, Section 211 FA 2013 (which regulates the proceedings before a
court or tribunal) even refers to the GAAR Guidance. Section 211(2) FA 2013 thereby provides that in
determining any issue in connection with the GAAR, a court or tribunal “must take into account (a)
the GAAR Guidance […] and (b) any opinion of the GAAR Advisory Panel about the arrangements.”
Accordingly, the GAAR Guidance is given a weight that goes beyond that of other forms of HMRC
Guidance, which comes very close to attaining legislative weight.170
Section 207 FA contains several subjective tests: the main purpose test,171 the tax advantage test172
and the so-called double-reasonableness test173. The main purpose test examines whether the
taxpayer implemented the arrangement with the intention of obtaining a tax benefit. The term
“obtaining of a tax advantage was the main purpose, or one of the main purposes” reveal that not
every (random) purpose that is attributable to an arrangement will be relevant. According to the
GAAR Guidance, the expressions regarding the “main purpose” have to be given “their normal
meaning as ordinary English words”.174 Thus, something has the “main” purpose where the purpose
makes up for the ‘most important’, ‘principal’, ‘leading’ or ‘the largest part’. This sheds only so much
light on the definition that it is safe to say that it is not the ‘sole’ purpose that is required.175 As with
the PPT of Action 6 as discussed above, it is unclear which purpose of an arrangement is the main
purpose, is there a hierarchy between different main purposes and at which levels of this hierarchy
would the tax related purpose be placed (always on top?). The UK GAAR Guidance provides some
explanation as to how a “main purpose” can be deduced – where (i) “the arrangement would not
have been carried out at all were it not for the opportunity to obtain the tax advantage” or where (ii)
“any non-tax objective was secondary to the benefit of the tax advantage”.176 Moreover, the “one of
169 The HMRC GAAR Guidance is available online: https://www.gov.uk/government/publications/tax-avoidance-general-anti-abuse-rules (Feb 25, 2017).
170 See J. Freedman, “United Kingdom”, in M. Lang et al (eds), GAARs – A Key Element of Tax Systems in the Post-BEPS Tax World? (Amsterdam: IBFD, 2015) 752; D. Roxburgh, “General Anti-Abuse Rule (GAAR)” (2014), 54 European Taxation 2/3, 113-116.
171 See GAAR Guidance, C4.3.
172 See Section 208 FA 2013; GAAR Guidance, C2.2.
173 See GAAR Guidance, C5.10.1.
174 See GAAR Guidance, C3.4.
175 See M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 208.
176 See GAAR Guidance, C3.5.
27
the main purposes” test seeks to establish whether an arrangement that “would otherwise have
occurred has been reshaped, or has been entered into under different terms and conditions, in order
to change significantly the tax result that would otherwise have arisen, and where the desired tax
result is itself a substantial objective.”177
The UK GAAR however makes the finding of abuse also conditional on the presence of a “tax
advantage”.178 The criterion “tax advantage” thereby serves a twofold purpose. On the one hand,
the tax advantage is a separate requirement necessary for the application of the GAAR. Hereby, it
serves as an absolute factor used to imply the presence of a tax abuse. On the other hand, the
concept of the tax advantage is also inevitably intertwined with the subjectivity element. This is
because the actual finding of a tax advantage is the result of a comparison of the arrangement
actually implemented by the taxpayer with a fictitious arrangement that the taxpayer would have
implemented in the absence of the abusive arrangement.179 Section 208 FA 2013 determines that a
“tax advantage includes
a) relief or increased relief from tax,
b) repayment or increased repayment of tax,
c) avoidance or reduction of a charge to tax or an assessment to tax,
d) avoidance of a possible assessment to tax,
e) deferral of a payment of tax or advancement of a repayment of tax, and
f) avoidance of an obligation to deduct or account for tax.”
Section 208 FA 2013 seems to be exhaustive, at least at first sight. However, the GAAR Guidance
states that the definition “is inclusive (i.e., it is not necessarily exhaustive) and is intended to have a
very wide meaning”180. Furthermore, it states that Section 208 FA 2013 “is intended to cover any
form of tax benefit, for example: increasing deductions or losses; decreasing income or gains;
obtaining timing advantages; obtaining or increasing repayments of tax; or ensuring that a potential
tax charge does not arise or is reduced.”181 Hence, although the definition seems to be very broad,
there might indeed be tax advantages that are not explicitly listed therein which could spark
discussions and court proceedings between taxpayers and tax administrations. Compared to the PPT
in Action 6 though it has to be acknowledged that the UK GAAR at least tries to define the “tax
177 See GAAR Guidance, C3.6.
178 See Section 208 FA 2013.
179 See M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 213.
180 See GAAR Guidance, C2.2.
181 See GAAR Guidance, C2.2.
28
advantage” or “tax benefit” respectively. The finding of a “tax advantage” is necessarily linked to the
subjective element. This is because a tax advantage is not simply given, but has to be deduced from
comparing the tax results of the arrangement with the tax results of a fictitious arrangement that
would have been implemented in the absence of the abusive arrangement. The procedure to identify
this fictitious arrangement is completely obscure, but at least the UK GAAR Guidance does not
request that the comparator is always the one fictitious arrangement that would give rise to the
greatest tax liability.182
Finally, Section 207(2) FA 2013 asks whether the carrying out of the arrangements can be
“reasonably be regarded as a reasonable course of action” (the so-called “double-reasonableness
test”). The double-reasonableness test resembles the main purpose test. The difference is that the
main purpose test aims at ascertaining whether the taxpayer entered into an arrangement for the
purpose of obtaining an advantage. The double-reasonableness test is aimed at ascertaining whether
a reasonable third party would have also carried out the arrangements. Apparently, the judge is
required to consider the range of reasonable views that could be held in relation to the
arrangements. If the judge considers that the arrangements “could reasonably be regarded as a
reasonable course of action”, the GAAR would not apply.183
While the traditional “main purpose test” did not get much attention by earlier scientific research,
the UK GAAR was very negatively received by the tax community. Especially its subjectivity and the
shifting of the burden of proof to the detriment of the taxpayer were highly criticized.184 However,
long before the codification of the UK GAAR, the concept of the traditional UK “main purpose test”
found its way not only into several of UK’s tax treaties but also into the domestic tax law of New
Zealand. The New Zealand courts established to look at the purpose or effect of the arrangement
when evaluating it in the light of New Zealand’s general anti-avoidance rule.185 Thus, to constitute tax
avoidance, an arrangement must have a tax avoidance purpose or an effect (“tax advantage”) that is
more than merely incidental. Conversely, the pursuit of a valid commercial objective that incidentally
182 See GAAR Guidance, C2.5: “in ascertaining whether an advantage arises the actual tax position should be compared with another tax position. The appropriate comparator or alternative tax position will depend on the facts, but will usually derive from the arrangements that would have occurred absent the abusive tax purpose (which may include no arrangement at all). In situations where there is more than one alternative arrangement that might have been adopted if the taxpayer had not adopted an abusive arrangement then the appropriate comparator would be the transaction that the taxpayer would most likely have carried out. This might not be the arrangement that would give rise to the greatest tax liability”.
183 See GAAR Guidance, C5.10.2.
184 See M. Seiler, “GAARs and Judicial Anti-Avoidance in Germany, the UK and the EU”, 220 et seq; M. Lang, “BEPS Action 6: Introducing an Antiabuse Rule in Tax Treaties”, 74 Tax Notes Int’l 7, 655-664; J. Freedman, “United Kingdom”, in M. Lang et al (eds), GAARs – A Key Element of Tax Systems in the Post-BEPS Tax World? 752; M. Gammie, “Moral taxation, immoral avoidance – what role for the law?” (2013), 58 British Tax Review 4, 577-590.
185 See further Z. Prebble and J. Prebble, “Comparing the General Anti-Avoidance Rule of Income Tax Law with the Civil Law Doctrine of Abuse of Law” (2008), 62 Bulletin for International Taxation 4, 151-170.
29
results in a reduction in tax liability is not tax avoidance. The tax avoidance purpose test though is
seen as being objective.186 The test examines whether the parties to the transaction/arrangement
would have entered into the transaction/arrangement even in the absence of a tax advantage. If the
parties would have entered into it even in the absence of the tax advantage, it does not have a tax
avoidance purpose.187
Even if considered an objective test, the judgment whether a transaction/arrangement would contra-
factually not be entered into when it was linked to different tax outcomes, could almost never be
objective. The lack of objectivity of the “main purpose test“ is illustrated by two court decisions from
New Zealand by the same judge on the same underlying circumstances with different outcomes –
Case V20188 and Case W33189. Both cases concern a dentist who had been a partner in a partnership
but left the partnership and established a trading trust structure whereby he was employed by the
trust. The transaction itself had commercially valid objectives such as to protect assets and limit
personal liability but it also led to tax savings in various degrees. The tax saving in the tax year 1995
(Case V20) was minor and thus Judge Barber held that there was no tax avoidance because the tax
advantage was merely incidental (“the icing on the cake”). However, in the year 1996 (Case W33) the
tax saving was not only larger but substantial. The larger tax saving influenced Judge Barber to hold
that the tax avoidance in that year was more than merely incidental and the Court held that the
arrangement in the second case had a tax avoidance purpose.
4. Anti-Treaty-Abuse measures currently employed by Asian countries
This paper specifically focuses on the tax treaty policy relating to treaty shopping of eight Asia Pacific
jurisdictions. These jurisdictions (Australia, China, Hong Kong, Japan, Malaysia, Singapore, South
Korea and Taiwan) are a very diverse group with respect to their levels of economic development,
their legal origins, their socio-economics and their levels of integration in the global trade system.
Three of the jurisdictions are OECD Member states,190 which thus at least presumably follow the
186 See further Z. Prebble and J. Prebble, “Comparing the General Anti-Avoidance Rule of Income Tax Law with the Civil Law Doctrine of Abuse of Law”, 62 Bulletin for International Taxation 4, 151-170.
187 See further Z. Prebble and J. Prebble, “Comparing the General Anti-Avoidance Rule of Income Tax Law with the Civil Law Doctrine of Abuse of Law”, 62 Bulletin for International Taxation 4, 151-170.
188 (2002) 20 NZTC 10,233 (TRA); see further Z. Prebble and J. Prebble, “Comparing the General Anti-Avoidance Rule of Income Tax Law with the Civil Law Doctrine of Abuse of Law”, 62 Bulletin for International Taxation 4, 151-170.
189 (2004) 21 NZTC 11,321 (TRA); see further Z. Prebble and J. Prebble, “Comparing the General Anti-Avoidance Rule of Income Tax Law with the Civil Law Doctrine of Abuse of Law”, 62 Bulletin for International Taxation 4, 151-170.
190 Australia, Japan and South Korea.
30
OECD recommendations when negotiating tax treaties and model their treaties closely after the
OECD-MC. Another two jurisdictions191 belong to the BRICS group, which keeps gaining influence in
international tax legislation.192 The third group of jurisdictions contains three emerging or high-
income developing countries193 - Malaysia, Singapore and Taiwan. The eight jurisdictions selected are
at varying levels of tax sophistication – with Australia and Japan at one end of the spectrum with
Malaysia arguably at the other end. They all have negotiated tax treaties that include some form of
anti-treaty abuse rules. The following section will analyze these treaty networks and present the
different anti-treaty abuse mechanisms these jurisdictions have employed in their respective tax
treaties.
4.1. The OECD Member States – Australia, Japan and South Korea
4.1.1. Australia
Australia currently194 has 44 tax treaties with countries around the world. Geographically, most tax
treaty partners are located in Europe (22 tax treaties) and in the Asia Pacific Region (16 tax treaties).
Out of the group of eight jurisdictions analyzed in this paper, Australia has a tax treaty with all of
them except for Hong Kong. The first tax treaty Australia entered into was the 1946 treaty with the
U.K.
Though they are numerous, anti-treaty abuse measures are not included consistently throughout the
whole Australian treaty network. The measure used most often is the “main purpose test” or a
version thereof (“MPT”). The treaty network contains 10 provisions that limit or deny the benefits of
the treaty if obtaining the treaty benefits was a “main purpose” of the respective arrangement or
transaction. The majority of these MPT provisions however are only applicable to passive income
(dividends, interests, royalties)195 and are therefore usually found in the respective passive income
articles of the treaty.196 The treaty with the United Kingdom197 applies the MPT more broadly to
191 China and Hong Kong (although Hong Kong is a special administrative region of China and not an official part of the BRICS grouping, we consider it here one of the BRICS due to it being part of China).
192 See further E. Eberhartinger and M. Petutschnig, “The dissenting opinion of BRICS Practitioners on the BEPS Agenda” (2017), 32 Australian Tax Forum 1, 1-57.
193 See “Development status groupings and composition” published by the United Nations Conference on Trade and Development” (available for download at http://unctadstat.unctad.org/EN/Classifications.html (Feb 25, 2017).
194 As of Feb. 25, 2017.
195 See the treaties with Chile (Apr. 1, 1973), Finland (Jan. 1, 2008), Ireland (Jul. 1, 1984), Japan (Jan. 1, 2009), Mexico (Jan. 1, 2004), New Zealand (May 1, 2010), Norway (Jan. 1, 2008), and South Africa (Jan. 1, 2009).
196 Except for the treaty with Chile where it is provided for in a separate Article 27.
197 Applicable as of 1 July 2004.
31
cover Other Income in addition to the dividend/interest/royalties. The treaty with Switzerland as
amended by the 2013 Protocol198 applies the MPT provision to the whole treaty and thus scrutinizes
all treaty benefits accordingly.
Besides the “main purpose test” the anti-abuse mechanism used most often are subject-to-tax
clauses. There are eight tax treaties199 with subject-to-tax clauses and/or “Limitation of Relief” (LoR)
clauses. Subject-to-tax-clauses limit the tax relief granted by a contracting state to the amount that
has actually been taxed / subject-to-tax in the other state, while a LoR clause limits the amount or
percentage of tax relief in the source state to the amount or share of income which is actually
remitted to or paid in the residence state.200 These clauses are not effectively provisions to counter
treaty shopping but to prohibit double non-taxation by relating the treaty benefit in one state to the
tax treatment in the other state.
Limitation on Benefits clauses (as found in U.S. treaties and the U.S. Model Tax Treaty) are only
included in two treaties, the treaties with Japan201 and the United States202 The treaty with the
United States was first signed in 1982, when it was one of the first treaties to include a specific
provision for income derived by a trustee in its LoB clause. With a Protocol in 2001 the treaty was
amended and the LoB clause was modernized to resemble the 1996 U.S. Model Treaty LoB clause
with some minor changes in the detailed wording. The LoB provision in the treaty with Japan is
modeled after the 2006 US Model Treaty LoB clause. Altogether though the treaty with Japan
represents a unique case. The treaty is the only one in the Australian treaty network that contains
four different concepts of anti-abuse measures. It contains a U.S.-style LoB clause203 for the whole
treaty, a “main purpose test”204 for passive income, a subject-to-tax clause205 and a specific anti-
treaty-shopping rule206 for passive income. Whether these different but to some degree overlapping
anti-abuse measures are really necessary and how the interplay of these measures works remains to
be seen.
198 Applicable as of 1 January 2014.
199 See the treaties with Argentina (Jan. 1, 2000), Ireland (Jul. 1, 1984), Malaysia (Aug. 9, 2010), Malta (Jan. 1, 1986), Singapore (Dec. 22, 2010), Thailand (Jan. 1, 1990) and the UK (Jul.1, 2004).
200 See for example the treaty between Australia and Malaysia (Jan. 1, 1980) or the treaty between Australia and Singapore (Jul. 1, 1969).
201 Applicable as of 1 January 2009.
202 Applicable as of 1 July 2001.
203 Art 23 of the Australia-Japan Tax Treaty.
204 Art 10(11), 11(10), 12(8) of the Australia-Japan Tax Treaty.
205 Art 24 of the Australia-Japan Tax Treaty.
206 Art 10(10), 11(9), 12(7) of the Australia-Japan Tax Treaty.
32
Finally, a number of Australian tax treaties also contain specific anti-abuse or anti-double-non-
taxation rules. These very specific and often rather casuistic provisions target preferential tax
treatments207 of certain industries (such as banks or insurances208), certain entities209 or certain
transactions210. Some of these specific anti-abuse measures are coupled with other, more standard
anti-abuse measures. But as they could also stand alone to form the only anti-abuse measure in the
treaty no specific trend in the Australian treaty policy could be found in this regard.
An analysis of the historical development of the Australian Tax Treaty network and the emphasis the
Australian treaty negotiators put on Anti-Avoidance Measures shows a time trend towards more of
such measures. Since the year 2000 the total number of Anti-Avoidance Measures increases.
However, the frequency of Anti-Avoidance measures in Tax Treaties has peaked during the period
2000-2009 and is declining since. As addressed above, MPT is the preferred measure. Yet, it seems
again that the usage of that measure is declining over the recent decade.
Australia Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per
Treaty
LoB per Anti-
Avoidance Measure
before 1990 (first Treaty
in 1946) 44 8 0.18 4 0.50 1 0.13 0.25 0 0.00 0.00
1990-1999 10 14 1.40 6 0.43 1 0.07 0.17 0 0.00 0.00
2000-2009 10 15 1.50 13 0.87 7 0.47 0.54 2 0.13 0.15
after 2009 7 7 1.00 3 0.43 1 0.14 0.33 0 0.00 0.00
Total 71 44 0.62 26 0.59 10 0.23 0.38 2 0.05 0.08
Table 1 - Historical Development Treaty Network - Australia
4.1.2. Japan
Japan currently211 has 61 tax treaties with countries around the world, with Europe the geographic
region with the most treaty partners (29 tax treaties) followed closely by the Asia Pacific region (24
207 See Art 7 of the 2002 Protocol to the Australia-Malaysia Tax Treaty that provides for a limitation of treaty benefits for persons “entitled to a particular tax treatment […] which has been identified by an exchange of Letters between the Contracting States”.
208 See Art 23 of the Australia-Russia Tax Treaty.
209 For example, holding companies -- see Art 23 of the Australia-Russia Tax Treaty.
210 See Art 7(b) of the Protocol to the Australia-Mexico Tax Treaty which denies treaty benefits to back-to-back loan arrangements.
211 As of February 25, 2017.
33
tax treaties). Out of the group of eight jurisdictions analyzed in this paper, Japan has a tax treaty with
all of them except for Taiwan. Japan’s first tax treaty negotiated was with the United States which
became effective in 1955.
Exactly half (31) of the Japanese tax treaties do not contain an anti-abuse measure of any sort.
Amongst the 30 tax treaties no specific preference for one of the two measures proposed by Acton 6
could be distinguished. Japan rather employs a number of different anti-abuse measures and in
numerous tax treaties (especially the more recent ones) combines two or more of these measures.212
With respect to the two measures proposed by Action 6, Japan’s tax treaties with the main purpose
test-provisions (13213) outnumber those with LoB provisions (8214). Similar to the Australian tax treaty
practice, the MPT provisions are usually exclusively applicable to passive income and are thus
incorporated in the respective passive income articles modeled after Articles 10, 11 and 12 of the
OECD-MC. While the treaty with Mexico reduces the scope of the MPT to royalties and interest only,
the treaties with France and the United Kingdom expand the MPT to Other Income and the treaty
with Hong Kong expands it even further to Other Income and capital gains. The LoB provisions are
modeled after the U.S. model – however all of them, except for the treaty with the United States, are
limited in their scope to certain items of income and distribution rules. These limitations to the scope
of the LoB clauses though seem to be rather random and a consistent policy cannot be identified.
The LoB clauses in the treaties with the Netherlands, Switzerland and the U.K. only apply to passive
income, capital gains and Other Income, while the LoB clause in the treaty with Australia is only
applicable to business income, dividends, interests and capital gains similar to the respective clause
in the treaty with France which additionally applies to Other Income as well. The LoB clause in the
treaty with Sweden is limited to passive income and the one in the treaty with New Zealand is limited
to interests and capital gains.
The one anti-abuse measure used most often besides the “main purpose test” is a “Limitation of
Relief”. There are 13215 tax treaties which have such a clause incorporated in the text. There are
212 See the tax treaties between Japan and Australia (Jan. 1, 2009), France (Jan. 1, 1997), Netherlands (Jan. 1, 2012), New Zealand (Jan. 1, 2014), Portugal (Jan. 1, 2014), Sweden (Jan.1, 2015), Switzerland (Jan. 1, 2012), UK (Jan. 1, 2014), US (Jul. 1, 2004).
213 See the tax treaties between Japan and Australia (Jan. 1, 2009), France (Jan. 1, 1997), Hong Kong (Jan. 1, 2012), Mexico (Jan. 1, 1997), New Zealand (Jan. 1, 2014), Oman (Jan. 1, 2015), Portugal (Jan. 1, 2014), Qatar (Jan. 1, 1997), Saudi Arabia (Jan. 1, 2012), Sweden (Jan. 1, 2015), Switzerland (Jan. 1, 2012), United Arab Emirates (Jan. 1, 2015), UK (Jan. 1, 2014).
214 See the tax treaties between Japan and Australia (Jan. 1, 2009), France (Jan. 1, 1997), Netherlands (Jan. 1, 2012), New Zealand (Jan. 1, 2014), Sweden (Jan. 1, 2015), Switzerland (Jan. 1, 2012), UK (Jan. 1, 2014), US (Jul. 1, 2004).
215 See the tax treaties between Japan and Australia (Jan. 1, 2009), Brunei (Jan. 1, 2010), France (Jan. 1, 1997), Ireland (Jan. 1, 1974), Kuwait (Jan. 1, 2014), Netherlands (Jan. 1, 2012), New Zealand (Jan. 1, 2014), Pakistan (Jan. 1, 2009), Portugal (Jan. 1, 2014), Singapore (Jul. 14, 2010), Switzerland (Jan. 1, 2012), U.K. (Jan. 1, 2014), U.S. (Jul. 1, 2004).
34
further twelve216 tax treaties which include a general anti-abuse rule that stipulates that both
contracting states “will ensure that any exemption or reduced rate of tax granted […] shall not be
enjoyed by persons not entitled to such benefits.” The two contracting states declare themselves
responsible to the other contracting state to adhere to this principle – however it is unclear how this
would be enforced.
As with the Australian tax treaties the Japanese treaty network also contains a few, very exceptional
specific anti-abuse measures. The treaty with South Korea contains a highly subjective general anti-
abuse provision which denies all treaty benefits when “the competent authorities of the Contracting
States agree that the taking advantage of those provisions constitutes an abuse […]”217. The treaty
with Luxembourg contains a specific provision which denies treaty benefits to Luxembourg holding
companies established according to the 1929 Act on holding companies.218 Japan’s treaty with
Malaysia contains an anti-treaty shopping provision in the form of a physical presence / active
business test which denies treaty benefits if a company does not conduct “substantive activities
through a fixed facility”219 in the respective state. The treaty with Singapore contains a similar
provision.220 The treaty with South Africa contains a kind of “main purpose test” for becoming a
resident – treaty benefits being denied if the main purpose of becoming a resident in either of the
contracting states was to benefit from the treaty.221
The historical analysis of the development of the Japanese Tax Treaty network shows a significant
time trend towards more Anti-Avoidance Measures (see Table 2 below). Every new Tax Treaty that
was entered into during the period 2000-2009 had on average two distinct Anti-Avoidance measures.
While this relation declined slightly after 2009, Japan still maintained the highest Anti-Avoidance
measure to Tax Treaty ratio of all eight jurisdiction in the sample during that time (1.89 Anti-
Avoidance measures per Tax Treaty). Table 2 also shows a significant change in the preferred
method. During the early 2000s (2000-2009) Japan negotiated more LoB-clauses per Anti-Avoidance
measures (0.33) than MPT-clauses (0.22). In recent years however, this practice reversed as Japan
216 See the tax treaties between Japan and Canada (Jan. 1, 2001), Finland (Jan. 1, 1992), Kazakhstan (Jan. 1, 2010), Luxembourg (Aug. 18, 2013), Mexico (Jan. 1, 1997), Netherlands (Jan. 1, 2012), Norway (Jan. 1, 1993), South Africa (Jan. 1, 1998), Sweden (Jan. 1, 2015), Turkey (Jan. 1, 1995), U.S. (Jul. 1, 2004), Vietnam (Jan. 1, 1996).
217 See Art 3 of the 1999 Protocol to the tax treaty between Japan and South Korea.
218 See Art 25 of the Japan-Luxembourg tax treaty.
219 See Art 5(a) of the 1999 Protocol to the tax treaty between Japan and Malaysia.
220 See Art 22(2) of the Japan-Singapore tax treaty.
221 See Art 22 of the Japan-South Africa tax treaty.
35
concluded twice as many MPT-clauses per Anti-Avoidance measure (0.56) than LoB-clauses (0.28).
Overall, Japan employs MPT-clauses (LoB-clauses) in 21% (23%) of its Treaties.
Japan Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per
Treaty
LoB per Anti-
Avoidance Measure
before 1990 (first Treaty
in 1955) 35 23 0.66 1 0.04 0 0.00 0.00 0 0.00 0.00
1990-1999 10 11 1.10 8 0.73 1 0.09 0.13 0 0.00 0.00
2000-2009 10 9 0.90 18 2.00 2 0.22 0.11 3 0.33 0.17
after 2009 7 18 2.57 34 1.89 10 0.56 0.29 5 0.28 0.15
Total 62 61 0.98 61 1.00 13 0.21 0.21 8 0.13 0.13
Table 2 - Historical Development Treaty Network - Japan
4.1.3. South Korea
South Korea currently222 has the second largest tax treaty network of the jurisdictions analyzed in this
paper with 87 tax treaties. The treaty partners are located all around the world, with Europe the
geographic region with the most treaty partners (36 tax treaties) closely followed by the Asia Pacific
region (33 tax treaties). South Korea is also the jurisdiction in our sample with the most treaties with
South American countries (8 tax treaties) and second most with African counties (7 tax treaties). Out
of the group of eight jurisdictions analyzed in this paper, South Korea has a tax treaty with all of them
except for Taiwan. The first South Korean tax treaty that came into effect was the treaty with Japan
in 1970. Since then the treaty network expanded rapidly, especially between the early 1990s and
2010 when the number of tax treaties almost tripled.
Out of South Korea’s 87 tax treaties, 36 (41% of all South Korean treaties) contain an anti-treaty
abuse provision of some sort. Many of these 36 treaties even contain two or more anti-abuse
measures with different scopes of application. The most preferred measure is the “main purpose
test” being used in 23 of the 36 treaties. When included in the tax treaty, the MPT provision usually
covers the passive income articles (dividends, interests, royalties) and is mostly provided for in the
222 As of February 25, 2017.
36
respective distribution rules. However the treaties with Ecuador223, Kuwait224, Kyrgyzstan225 and Saudi
Arabia226 expand the main purpose test onto the whole treaty and all its distribution rules. In the
majority of the other treaties that contain a main purpose test this test applies to passive income,
capital gains and Other Income.227 The treaties with Germany228, Oman229 and the United Kingdom230
apply the test to passive income and Other Income. A fourth group of treaties on the other hand
reduce the scope of the main purpose test to either interests and royalties,231 or only interests,232 or
only royalties.233 A special case is the tax treaty with Pakistan as it provides for a main purpose test
which exclusively applies to business income, i.e., the attribution of profits to permanent
establishments.234
The other anti-abuse measures in South Korea’s tax treaties are mainly anti-treaty shopping rules
that employ certain aspects (predominantly the ownership test) of the standard U.S. LoB clause.
These anti-treaty shopping rules usually deny treaty benefits with respect to passive income, capital
gains and Other Income to intermediaries, which are directly or indirectly controlled by residents of
neither of the contracting states.235 This rule is a very mechanical test that denies treaty benefits
exclusively on the grounds of ownership by a third state resident. A safe harbor or bona fide rule
such as, for example, an active trade or business test is only very rarely added to the mechanical
test.236 A traditional, comprehensive LoB clause cannot be found in the South Korean tax treaty
network. The treaty with the United Arab Emirates contains a limitation on benefits clause – however
its scope is so broad that it limits the application of the treaty for residents of the United Arab
Emirates to either individuals or government owned companies.237 Additionally, this LoB clause is
223 See Art 25(1)(b) of the South Korea-Ecuador tax treaty (Jan. 1, 2014).
224 See Art 6 of the 2011 Protocol to the South Korea-Kuwait tax treaty (Jan. 1, 2011).
225 See Art 29 of the South Korea-Kyrgyzstan tax treaty (Jan. 1, 2014).
226 See Art 27 of the South Korea-Saudi Arabia tax treaty (Jan. 1, 2009).
227 See the tax treaties between South Korea and Azerbaijan (Jan. 1, 2009), Bahrain (Jan. 1, 2014), Colombia (Jan. 1, 2015), Hong Kong (Jan. 1, 2017)), Panama (Jan. 1, 2013), Peru (Jan. 1, 2015), Poland (Protocol 2013 – Jan. 1, 2015), Qatar (Jan. 1, 2010) and Uruguay (Jan. 1, 2014).
228 See Art 27(2) of the South Korea-Germany tax treaty (Jan. 1, 2003).
229 See Art 10(6), 11(9), 12(7) and 22(3) of the South Korea-Oman tax treaty (Jan. 1, 2007).
230 See Art 10(6), 11(10), 12(7) and 22(4) of the South Korea-United Kingdom tax treaty (Jan. 1, 1997).
231 See the tax treaties between South Korea and Chile (Jan. 1, 2004), Mexico (Jan. 1, 1996) and Ukraine (Jan. 1, 2003).
232 See Art 11(9) of the South Korea-Papua New Guinea tax treaty (Jan. 1, 1999).
233 See Art 12(7) of the South Korea-Uzbekistan tax treaty (Jan. 1, 1999).
234 See Art 2 of the Protocol to the South Korea-Pakistan tax treaty (Jan. 1, 1987).
235 See the tax treaties between South Korea and Bahrain (Jan. 1, 2014), Canada (Jan. 1, 2007), China (Protocol 2006 – Jul. 4, 2016), Colombia (Jan. 1, 2015), Ecuador (Jan. 1, 2014), Kuwait (Jan. 1, 2011), Latvia (Jan. 1, 2010), Panama (Jan. 1, 2013), Thailand (Jan. 1, 2008), Turkey (Jan. 1, 1987), United States (Jan. 1, 1979) and Uruguay (Jan. 1, 2014).
236 See for example Art 1 of the 2006 Protocol (Jul. 4, 2006) to the South Korea-China tax treaty.
237 See Art 23(1) and (2) of the South Korea-United Arab Emirates tax treaty (Jan. 1, 2003).
37
complemented by a main purpose test that further limits the application of the treaty for UAE
residents.238
South Korea Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per
Treaty
LoB per Anti-
Avoidance Measure
before 1990 (first Treaty
in 1970) 20 17 0.85 6 0.35 1 0.06 0.17 0 0.00 0.00
1990-1999 10 27 2.70 8 0.30 4 0.15 0.50 0 0.00 0.00
2000-2009 10 28 2.80 16 0.57 7 0.25 0.44 1 0.04 0.06
after 2009 7 15 2.14 21 1.40 11 0.73 0.52 0 0.00 0.00
Total 47 87 1.85 51 0.59 23 0.26 0.45 1 0.01 0.02
Table 3 - Historical Development Treaty Network - South Korea
Analyzing the historical development of the South Korean Tax Treaty network in detail provides a
consistent and significant trend towards a more pronounced use of Anti-Avoidance Measures over
time. During all of the four periods presented in Table 3 the total number of Anti-Avoidance
measures increases (the frequency increases in three of the four eras). The most recent period (after
2009) is characterized by a strong increase (in total and relative numbers) of Anti-Avoidance
measures in general and more specifically of MPT-clauses. Every new Treaty has on average 1.4 Anti-
Avoidance measures of which 52% are MPT-clauses. The total number of MPT-clauses almost
doubled after 2009 and their usage in new Treaties almost tripled to the previous period (2000-
2009). As addressed above, LoB-clauses do not matter at all.
4.2. China and Hong Kong
4.2.1. China
With 99 tax treaties China currently239 has the largest tax treaty network of the jurisdictions analyzed
in this paper. The treaty partners are located all around the world, with Europe the geographic region
with the most treaty partners (40 tax treaties) closely followed by the Asia Pacific region (38 tax
treaties). China is also the jurisdiction in our sample with the most treaties with African countries (11
238 See Art 23(3) of the South Korea-United Arab Emirates tax treaty (Jan. 1, 2003).
239 As of February 25, 2017.
38
tax treaties). From the group of eight jurisdictions analyzed in this paper, China has a tax treaty with
all of them except for Taiwan. Even if China has the highest number of tax treaties currently in force,
it has a rather short history of tax treaty negotiations as its first treaties (with Germany, Japan and
the United Kingdom, respectively) did not became effective until 1985.
Only 24 of the 99 tax treaties contain some form of anti-treaty abuse provisions. Similar to South
Korea, the most preferred measure is the main purpose test which is included in 80% (19 tax treaties)
of the tax treaties that have some form of an anti-abuse provision. Using the main purpose test
however seems to be a very new feature of the Chinese tax treaty policy since almost all treaties240
that contain a MPT provision have either been first signed or amended by a protocol after 2009. It is
also a very persistent treaty policy since 2010. Since then only a handful of tax treaties, especially
with developing countries, have been concluded that do not contain MPT provisions.241 The main
purpose test in China’s tax treaties is usually reduced in scope to passive income242 or passive income
and Other Income243 and therefore is stipulated in the respective treaty articles dealing with
dividends, interests, royalties and Other Income. In some cases the MPT provision is applicable to the
whole treaty or all distribution rules respectively,244 while only the treaty with Papua New Guinea
further reduces the scope of the main purpose test to interests only.245
China’s treaty network contains four treaties which have a limitation on benefits clause. All of these
LoB clauses are modeled after the U.S. model LoB clause or a modification thereof.246 They all use
ownership and base erosion tests as well as safe harbor provisions for certain types of organizations
such as not-for-profit organizations or publicly traded companies. The treaties with Mexico, Russia
and the United States supplement the LoB clause with a main purpose test for the entire treaty
(Mexico) or for passive income and Other Income (Russia, United States). The treaty with Ecuador is
240 The treaty with Papua New Guinea went into effect in 1996, the treaty with New Zealand was amended accordingly in 2001 and the treaty with Mexico entered into force in 2006.
241 See the tax treaties between China and Ethiopia (Jan. 1, 2013), Nepal (Jan. 1, 2011), Syria (Jan. 1, 2012), Turkmenistan (Jan. 1, 2011), Uganda (signed Jan. 11, 2012 – still pending) and Zambia (Jan. 1, 2012). The tax treaty with Ecuador (Jan. 1, 2015) contains an LoB clause.
242 See the tax treaties between China and Belgium (Jan. 1, 2014), Botswana (signed Apr. 2012 – still pending), Finland (Jan. 1, 2011), Nigeria (Jan. 1, 2010), and Singapore (Jan. 1, 2011).
243 See the tax treaties between China and Denmark (Jan. 1, 2013), France (Jan. 1Jan. 1, 2015), Malta (Jan. 1, 2012), Russia (Jan. 1, 2017), Switzerland (Jan. 1, 2015), and the United Kingdom (Jan. 1, 2014).
244 See the tax treaties between China and Czech Republic (May 4, 2011), Germany (Jan. 1, 2017), Mexico (Mar. 1, 2006), New Zealand (Apr. 1, 2001) and the United States (Nov. 24, 2010).
245 See Art 11(8) of the China-Papua New Guinea tax treaty (Jan. 1, 1996).
246 See the tax treaties between China and Ecuador (Jan. 1, 2015), Mexico (Mar. 1, 2006), Russia (Jan. 1, 2017) and the United States (Nov. 24, 2010).
39
the only Chinese tax treaty that exclusively employs a LoB clause as the sole anti-treaty shopping
mechanism.
As with the other jurisdictions’ tax treaty networks, China’s treaty network also contains a number of
additional anti-abuse measures. One repeatedly used clause explicitly preserves for both contracting
states the right to counter treaty-based tax abuse with their domestic tax laws and declares domestic
anti-abuse rules as conforming with the tax treaty even if such rules would otherwise constitute a
treaty override.247 The treaty with the Czech Republic contains a highly subjective general anti-abuse
provision similar to the one found in Article 3 of the 1999 Protocol to the tax treaty between Japan
and South Korea, which denies all treaty benefits when “the competent authorities of the Contracting
States agree that the taking advantage of those provisions constitutes an abuse […]”248. A similar
provision can also be found in Article 4 of the Protocol to the tax treaty between China and Israel.
Article V of the Protocol to the China-Mexico tax treaty contains a provision denying treaty benefits
in cases of thin-capitalization, CFC rules and back-to-back loan arrangements.
China Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per
Treaty
LoB per Anti-
Avoidance Measure
before 1990 (first Treaty
in 1985) 5 8 1.60 0 0.00 0 0.00 n/a 0 0.00 n/a
1990-1999 10 30 3.00 4 0.13 1 0.03 0.25 0 0.00 0.00
2000-2009 10 34 3.40 7 0.21 2 0.06 0.29 1 0.03 0.14
after 2009 7 27 3.86 32 1.19 16 0.59 0.50 3 0.11 0.09
Total 32 99 3.09 43 0.43 19 0.19 0.44 4 0.04 0.09
Table 4 - Historical Development Treaty Network - China
China’s Tax Treaty network expanded very quickly since 1990 with more than three new Treaties per
year and has thus the fastest growing Treaty network over that time span (only the growth rates of
Singapore and Hong Kong since 2009 are bigger). The strongest increase of Anti-Avoidance measures
can be found during the most recent period (after 2009) with 32 total and 1.19 Anti-Avoidance
measures per new Treaty. Table 4 also shows a strong propensity of negotiating more MPT-clauses in
recent years. 50% of the new Anti-Avoidance measures are MPTs and almost 60% of the new
247 See for example Art 23 of the China-Denmark tax treaty.
248 See Art 21(3) of the China-Czech Republic tax treaty.
40
Treaties contain such a clause. At a very low total, yet still significant, China employs more LoB
clauses since 2009.
4.2.2. Hong Kong
Hong Kong has a very brief history in tax treaty negotiations as the first treaty came into effect on
April 1, 2004 (tax treaty with Belgium). Given this, the current overall number of 32 tax treaties249
with countries all over the world250 is quite impressive. Hong Kong has 17 European tax treaty
partners, 12 from the Asia Pacific region, treaties with Canada and Mexico in North America and a tax
treaty with one country in Africa (South Africa). Out of the group of eight jurisdictions analyzed in
this paper, Hong Kong has a tax treaty (or tax treaty equivalent) with China, Japan, Malaysia and
South Korea.
Hong Kong’s tax treaties in large part contain measures to curb tax abuse and treaty shopping. Only 5
of the 32 tax treaties (15%) do not contain any anti-abuse rules.251 Twenty tax treaties contain a
general anti-abuse rule which preserves the right of each party to the treaty “to apply its domestic
laws and measures concerning tax avoidance”252. In many of its treaties, these general anti-abuse
provisions are further supplemented with more specific anti abuse rules. Some treaties however also
include either only the above-mentioned general rule or only a more specific provision. With regards
to the specific anti-treaty shopping rules, the main purpose test is again the most commonly used.
Thirteen treaties contain such a provision. Six of these thirteen treaties limit the application of the
respective MPT provisions to passive income,253 four treaties limit it to passive income, capital gains
and Other Income,254 while the treaties with France255 and the United Kingdom256 limit the scope to
passive income and capital gains or passive income and Other Income, respectively. A further
exception is the treaty with Brunei which applies the main purpose test on technical service fees and
on passive income.257 In addition, Hong Kong’s tax treaties contain several other anti-abuse rules –
249 As of February 25, 2017.
250 Hong Kong has no tax treaty with a country in South America.
251 See the tax treaties between Hong Kong and Austria (Apr. 1, 2012), Hungary (Apr. 1, 2012), Ireland (Apr. 1, 2012), Liechtenstein (Apr. 1, 2012) and Malaysia (Apr. 1, 2013).
252 See e.g., Art 27 of the Hong Kong-Belgium tax treaty.
253 See the tax treaties between Hong Kong and Canada (Apr. 1, 2014), Indonesia (Apr. 1, 2012), Italy (Aug. 14, 2015), New Zealand (Apr. 1, 2012), Qatar (Apr. 1, 2014) and South Africa (Oct. 20, 2015).
254 See the tax treaties between Hong Kong and Japan (Aug. 14, 2011), South Korea (Sept. 27, 2016), Portugal (Apr. 1, 2013) and Spain (Apr. 1, 2012).
255 See Art 10(6), 11(8), 12(7) and 13(6) of the Hong Kong-France tax treaty (Apr. 1, 2012).
256 See Art 10(6), 11(7), 12(7) and 20(6) of the Hong Kong-U.K. tax treaty (Apr. 1, 2011).
257 See Art 10(5), 11(8), 12(7) and 13(7) of the Hong Kong-Brunei tax treaty (Apr. 1, 2011).
41
for example, applying an ownership test to deny treaty benefits of companies held/controlled by
third country residents.258 None of Hong Kong’s treaties contain a LoB-clause.
Hong Kong Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per
Treaty
LoB per Anti-
Avoidance Measure
2000-2009 (first Treaty
in 2004) 6 3 0.50 3 1.00 0 0.00 0.00 0 0.00 0.00
after 2009 7 29 4.14 33 1.14 13 0.45 0.39 0 0.00 0.00
Total 13 32 2.46 36 1.13 13 0.41 0.36 0 0.00 0.00
Table 5 - Historical Development Treaty Network - Hong Kong
Hong Kong has a very short history of sovereign Tax Treaty negotiations; however, this short history
is characterized by a very strong emphasis on Anti-Avoidance measures. On average every Treaty
contains of 1.13 Anti-Avoidance measures. While MPT provisions were not used during Hong Kong’s
first six years of Treaty negotiations (2004-2009), since 2009 45% of new Treaties include a MPT-
clause (13 of 29 Treaties).
4.3. Three Emerging Economies – Malaysia, Singapore, Taiwan
4.3.1. Malaysia
Malaysia has the fourth largest tax treaty network of the eight jurisdictions in our study with 73
treaties around the world.259 It has a treaty with all other jurisdictions in our study. The largest
number of treaty partners by region are in the Asia Pacific (36 tax treaties) followed by Europe (27
tax treaties). However, it does not have a treaty with the United States. The first ever treaty was
signed in 1962 (effective August 30, 1963) with the United Kingdom. Almost 70% of Malaysia’s tax
treaties (50 treaties) do not contain an anti-abuse provision of any sort. Of the treaties that do, most
often employed are Limitation of Relief (“LoR”) clauses, followed by main purpose tests and subject-
to-tax clauses. Traditional LoB clauses in the sense of Art X(1)(5) of Action 6 are nowhere to be found
in the current Malaysian tax treaty network.
258 See Art 26(3) of the Hong Kong-Canada tax treaty (Apr. 1, 2014), Art 9 of the Protocol to the Hong Kong-Mexico tax treaty (Apr. 1, 2014), Art 10(8), 11(5), 12(7) and 21(3) of the Hong Kong-Switzerland tax treaty (Apr. 1, 2013).
259 As of February 25, 2017.
42
The most often employed LoR clause limits the amount or percentage of treaty-based tax relief in the
source state to the amount or share of income which is actually remitted to or paid in the residence
state.260 Six of these LoR clauses are additionally intertwined with a subject-to-tax clause that denies
the relief altogether if the income had not been taxed or properly declared and assessed in the other
contracting state.261
The main purpose test, which denies treaty benefits when the main purpose or one of the main
purposes of an arrangement or a transaction was to take advantage of the respective treaty, is
provided for in seven of Malaysia’s tax treaties. A consistent treaty policy with regards to the scope
of application of these MPT provisions however cannot be observed. The treaty with India applies
the MPT to all income / distribution rules of the treaty.262 The treaties with Chile263 and the United
Kingdom264 apply the test to all items of passive income, while the treaty with Kazakhstan265 limits
the application to royalties and interests. The narrowest scope of MPT application – interests only –
is contained in the treaties with New Zealand266 and Papua New Guinea267. In the Malaysia-Poland
treaty, the MPT applies to passive income and technical service fees.268
Besides the more often used LoR clauses and the MPT, Malaysia’s tax treaties contain several very
specific anti-abuse measures. The treaties with India and Spain contain an active trade or business
test.269 The treaty with Spain also provides for an anti-treaty shopping measure denying treaty
benefits with respect to passive income and capital gains to companies (in)directly held/controlled by
third country residents.270 And, as mentioned above, the treaty with Japan employs a substantive
physical presence test, which denies treaty benefits if a company does not conduct “substantive
activities through a fixed facility”271 in the respective state.
260 The 17 tax treaties which employ such a Limitation of Relief clause are the treaties with: Australia (Jan. 1, 1980), Canada (Jan. 1, 1981), Denmark (Protocol -- Jan. 1, 2003), Germany (Protocol -- Jan. 1., 2011), Ireland (Protocol -- Jan. 1, 2012), Italy (Jan. 1, 1977), Luxembourg (Jan. 1, 2005), Myanmar (Jan. 1, 2009), Netherlands (Protocol -- Jan. 1, 2011), New Zealand (Protocol –Jan. 12, 2016), Pakistan (Jan. 1, 1980), Papua New Guinea (Jan. 1, 2000), South Korea (Jan. 1, 1983), Sweden (Jan. 1, 2006), Taiwan (Jan. 1, 2000), Thailand (Jan. 1, 1983), United Kingdom (Jan. 1, 2000).
261 See the treaties between Malaysia and Australia (Jan. 1, 1980), Canada (Jan. 1, 1981), Germany (Protocol -- Jan. 1, 2011), Netherlands (Protocol -- Jan. 1, 2011), New Zealand (Protocol –Jan. 12, 2016), Pakistan (Jan. 1, 1980).
262 See Art 28(2) of the Malaysia-India tax treaty (Jan. 1, 2013).
263 See Art 10(6), 11(7) and 12(7) of the Malaysia-Chile tax treaty (Jan. 1, 2009).
264 See Art 10(6), 11(7) and 12(7) of the Malaysia-U.K. tax treaty (Jan. 1, 2000).
265 See Art 11(8) and 12(7) of the Malaysia-Kazakhstan tax treaty (Jan. 1, 2011).
266 See Art 1(a) of the 1994 Protocol to the Malaysia-New Zealand tax treaty (Jul. 1, 1996).
267 See Art 11(10) of the Malaysia-Papua New Guinea tax treaty (Jan. 1, 2000).
268 See Art 10(6), 11(8), 12(7) and 13(7) of the Malaysia-Poland tax treaty (Protocol – Jan. 1, 2015).
269 See Art 28(3) of the Malaysia-India tax treaty (Jan. 1, 2013), Art 5(b) of the Protocol to the Malaysia-Spain tax treaty (Jan. 1, 2008).
270 See Art 5(a) of the Protocol to the Malaysia-Spain tax treaty (Jan. 1, 2008).
271 See Art 5(a) of the 1999 Protocol to the Malaysia-Japan tax treaty (Jan. 1, 2000).
43
Malaysia Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per
Treaty
LoB per Anti-
Avoidance Measure
before 1990 (first Treaty
in 1963) 27 17 0.63 4 0.24 0 0.00 0.00 0 0.00 0.00
1990-1999 10 11 1.10 1 0.09 1 0.09 1.00 0 0.00 0.00
2000-2009 10 30 3.00 10 0.33 2 0.07 0.20 0 0.00 0.00
after 2009 7 15 2.14 12 0.80 4 0.27 0.33 0 0.00 0.00
Total 54 73 1.35 27 0.37 7 0.10 0.26 0 0.00 0.00
Table 6 - Historical Development Treaty Network - Malaysia
Malaysia’s Tax Treaty network contains the lowest ratio of Anti-Avoidance measures per Tax Treaty
of the whole sample with 37%. Table 6 also shows Malaysia’s reluctance of negotiating MPTs. Only
seven Treaties (less than 10%) contain such a clause. Notwithstanding the small total number of
MPTs, its usage seems to be increasing over the last 7 years. The number of Treaties containing such
a provisions more than doubled after 2009 and also the MPT-to-Anti-Avoidance-measure ratio
increased by more than 50% since 2009.
4.3.2. Singapore
Singapore currently272 has the third largest tax treaty network of the jurisdictions analyzed in this
paper with 82 tax treaties, which is remarkable considering Singapore’s size and population. The
treaty partners are located all around the world, with Europe being the geographic region with the
most treaty partners (40 tax treaties) followed by the Asia Pacific region (31 tax treaties). Singapore
also has a tax treaty with all of the jurisdictions analyzed in more depth in this paper, except for Hong
Kong. However, Singapore like Malaysia does not have a treaty with the United States. The first tax
treaties concluded by Singapore were those with Malaysia, Norway and the United Kingdom, all in
1966.
Singapore’s anti-treaty abuse policy is straightforward and consistent. It rests on two pillars:
Limitation on Relief provisions (in 39 tax treaties) and main purpose test provisions (in 18 tax
treaties). While more than 58% of Singapore’s treaties contain some sort of anti-abuse rules, 34 tax
272 As of February 25, 2017.
44
treaties employ no anti-treaty abuse provision. The treaties which contain anti-abuse rules very often
contain multiple provisions – mostly Limitation of Relief rules and main purpose test provisions, but
sometimes also a general anti-avoidance rule. Limitation on benefits clauses are not employed by
any of Singapore’s existing 82 tax treaties.
The scopes of application of the main purpose test provisions in Singapore’s tax treaties are rather
diverse: six treaties apply the test to the whole treaty,273 five treaties apply it exclusively to passive
income,274 three treaties reduce the application of the test to the distribution rules for interests and
royalties,275 and the treaty with New Zealand276 reduces the scope of the test to dividends and
royalties. The narrowest scope of application however is the MPT provision in the treaty with Brunei
where it is only applicable to interests.277
Among the other miscellaneous anti-abuse rules in Singapore’s tax treaty network, a few are
especially noteworthy – five treaties contain a general anti-avoidance clause, which preserves the
right of each party to the treaty “to apply its domestic laws and measures concerning tax
avoidance”278. The treaty with Japan279 contains an anti-treaty shopping provision in the form of a
physical presence / active business test which denies treaty benefits if a company does not conduct
“substantive activities through a fixed facility” 280 in the respective state. The treaty with India
contains an anti-conduit rule, that applies a form of the active trade or business test and denies
treaty benefits to resident companies with “negligible or nil business operations or with no real and
continuous business activities”281 carried out in the residence state. Finally, the treaty with Mauritius
employs an anti-harmful tax competition rule under which the Article 10, 11 and 12 benefits are not
applicable with respect to companies subject to preferential tax treatments in Mauritius.282
273 See Article 27 of the Singapore-Ecuador tax treaty (Jan. 1, 2016); Art 22(3) of the Singapore-Estonia tax treaty (Jan. 1, 2009); Art 22(3) of the Singapore-Finland tax treaty (Apr. 30, 2010); Art 28 of the Singapore-France tax treaty (Jun. 1, 2016); Art 3(1) of the 2005 Protocol to the Singapore-India tax treaty (Jan. 1, 2008); Art 22(3) of the Singapore-Latvia tax treaty (Jan. 1, 2002).
274 See Art 10(6), 11(8), 12(7) of the Singapore-China tax treaty (Jan. 1, 2008); Art 10(8), 11(9), 12(7) of the Singapore-Oman tax treaty (Jan. 1, 2008); Art 10(8), 11(8), 12(7) of the Singapore-Poland tax treaty (Jan. 1, 2015) – note: according to Art 4(d) of the 2005 Protocol to the Singapore-Poland treaty, the MPT is also applicable to the tax sparing clause; Art 1(d) of the 2011 Protocol to the Singapore-Spain tax treaty (Jan. 1, 2013); Art 10(7), 11(9), 12(8) of the Singapore-U.K. tax treaty (Jan. 1, 1999).
275 See Art 11(8), 12(7) of the Singapore-Mexico tax treaty (Jan. 1, 1996); Art 11(10), 12(7) of the Singapore-Romania tax treaty (Jan. 1, 2003); Art 11(8), 12(7) of the Singapore-Ukraine tax treaty (Jan. 1, 2010).
276 See Art 10(6), 12(7) of the Singapore-New Zealand tax treaty (Jan. 1, 2011).
277 See Art 11(8) of the Singapore-Brunei tax treaty (Jan. 1, 2007).
278 See Art 26 of the Singapore-China tax treaty (Jan. 1, 2008); Art 11(b) of the 2013 Protocol to the Singapore-Czech Republic tax treaty (Jan. 1, 2016); Art 27 of the Singapore-Ecuador tax treaty (Jan. 1, 2016); Art 9(c)(i) of the 2005 Protocol to the Singapore-Israel tax treaty (Jan. 1, 2007); Art 1(a)-(c) of the 2011 Protocol to the Singapore-Spain tax treaty (Jan. 1, 2013).
279 See also above.
280 See Art 22(2) of the Japan-Singapore tax treaty.
281 See Art 3(1)-(4) of the 2005 Protocol to the Singapore-India tax treaty (Jan. 1, 2008).
282 See Art 2 of the 1995 Protocol to the Singapore-Mauritius tax treaty (Jan. 1, 1997).
45
Singapore Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per Treaty
LoB per Anti-
Avoidance Measure
before 1990 (first Treaty
in 1966) 24 5 0.21 5 1.00 0 0.00 0.00 0 0.00 0.00
1990-1999 10 7 0.70 9 1.29 1 0.14 0.11 0 0.00 0.00
2000-2009 10 18 1.80 15 0.83 6 0.33 0.40 0 0.00 0.00
after 2009 7 52 7.43 34 0.65 11 0.21 0.32 0 0.00 0.00
Total 51 82 1.61 63 0.77 18 0.22 0.29 0 0.00 0.00
Table 7 - Historical Development Treaty Network - Singapore
Singapore expanded its Tax Treaty network after 2009 by more than seven Treaties per year. The
emphasis on Anti-Avoidance measures however seems to have declined but is still at a high level.
65% of those new Tax Treaties contain an Anti-Avoidance measure, which is a reduction by 18
percentage points to the previous period (2000-2009). Table 7 also presents an increase of the total
number of MPT-clauses after 2009. Again this increase is not as pronounced as it was during the
early 2000s (2000-2009).
4.3.3. Taiwan
The smallest tax treaty network of the jurisdictions in our study is that of Taiwan. Taiwan currently283
has 28 tax treaties with thirteen European, ten Asia Pacific, four African and one South American
country. Within our sample of eight Asia Pacific jurisdictions, Taiwan has tax treaties with Australia,
Malaysia and Singapore, which was also the first country to conclude a tax treaty with Taiwan in
1982.
16 tax treaties contain some sort of anti-abuse rule, with the main purpose test being the
predominantly used anti-abuse measure (10 tax treaties). Three treaties contain limitation of relief
clauses which limit the amount or percentage of tax relief in the source state to the amount or share
of income which is actually remitted to or paid in the residence state.284 Interestingly, there is one
treaty that employs a limitation on benefits clause.285 In all of the treaties that contain the MPT, it is
283 As of February 25, 2017.
284 See Art 22 of the Taiwan-Malaysia tax treaty (Jan. 1, 2000); Art 17 of the Taiwan-Singapore tax treaty (Jan. 1, 1982); Art 23(1) of the Taiwan-U.K. tax treaty (Dec. 23, 2002).
285 See Art 26(1) of the Taiwan-Denmark tax treaty (Jan. 1, 2006).
46
applied to the entire treaty,286 with the sole exception of the treaty with the United Kingdom287
wherein the MPT is applicable only to passive income. Further, the treaties with Denmark288 and
Sweden289 contain subject-to-tax clauses for income in connection to certain activities (e.g., finance,
headquarters activities, etc.). The treaty with India contains an anti-conduit rule, based on an active
trade or business test.290
Taiwan Years Number of Tax
Treaties
Treaties per
Year
Anti-Avoidance Measures
(Total)
Anti-Avoidance Measures per Treaty
MPT MPT per
Treaty
MPT per Anti-
Avoidance Measure
LoB LoB per
Treaty
LoB per Anti-
Avoidance Measure
before 1990 (first Treaty
in 1982) 8 1 0.13 1 1.00 0 0.00 0.00 0 0.00 0.00
1990-1999 10 8 0.80 0 0.00 0 0.00 n/a 0 0.00 n/a
2000-2009 10 8 0.80 9 1.13 2 0.25 0.22 1 0.13 0.11
after 2009 7 11 1.57 18 1.64 8 0.73 0.44 0 0.00 0.00
Total 35 28 0.80 28 1.00 10 0.36 0.36 1 0.04 0.04
Table 8 - Historical Development Treaty Network - Taiwan
Taiwan started to consistently negotiate Anti-Avoidance measures in 2000. Since then every new
Treaty contains on average of 1.38 Anti-Avoidance rules. This rate has been steadily increasing and
ranges at 1.64 Anti-Avoidance measures per Treaty by 2016. While the results in Table 8 do not show
any preference between MPTs and LoBs during the early 2000s (2000-2009), the trend is quite clear
after 2009. The usage of MPTs increased consistently in total numbers, per new Treaty and per Anti-
Avoidance measure after 2009.
286 See Art 25 of the Taiwan-Austria tax treaty (Jan. 1, 2015); Art 27 of the Taiwan-Belgium tax treaty (Jan. 1, 2006); Art 26(3) of the Taiwan-Denmark tax treaty (Jan. 1, 2006); Art 27(1) of the Taiwan-France tax treaty (Jan. 1, 2011); Art 26 of the Taiwan-Hungary tax treaty (Jan. 1, 2011); Art 28(1) of the Taiwan-India tax treaty (Aug. 12, 2011); Art 27(1) of the Taiwan-Israel tax treaty (Jan. 1, 2010); Art 26 of the Taiwan-Kiribati tax treaty (Jan. 1, 2015); Art 27 of the Taiwan-Luxembourg tax treaty (Jan. 1, 2015); Art 26(1) of the Taiwan-Slovakia tax treaty (Jan. 1, 2012); Art 26(3) of the Taiwan-Switzerland tax treaty (Dec. 13, 2011).
287 See Art 10(6), 11(7), 12(7) and 23(1) of the Taiwan-U.K. tax treaty (Dec. 23, 2002).
288 See Art 26(1) of the Taiwan-Denmark tax treaty (Jan. 1, 2006).
289 See Art 26 of the Taiwan-Sweden tax treaty (Nov. 24, 2004).
290 See Art 28(2) of the Taiwan-India tax treaty (Aug. 12, 2011).
47
5. Going Forward – Some policy implications
The analysis of the tax treaty networks of eight Asia Pacific jurisdictions provides a timely snapshot of
the tax treaty abuse policy positions of these jurisdictions (see Table 10 below for an overview).291
Overall, the eight jurisdictions have a total of 506 tax treaties currently in force or signed but
ratification still pending. On average this is more than 63 tax treaties per jurisdiction. China has the
largest tax treaty network with 99 treaties, followed by South Korea (87) and Singapore (82), with
Taiwan having the smallest treaty network (28 treaties). Geographically the treaty networks are truly
global. Overall European countries are most often the “other contracting state” (with 224 treaties)
followed by Asia Pacific (200), the Americas (43) and Africa (29). Malaysia is the only country in the
sample that has more treaties with other Asia Pacific jurisdictions (36) than with European countries
(27). The majority of the tax treaties are modeled after the OECD-MC especially those concluded with
OECD member states. Elements of the UN Model Convention can also be found in some of the
treaties, in particular those between developing countries as well as rather autochthone rules.
Jurisdiction Europe Asia Pacific North
America
South
America Africa
Australia 22 16 3 2 1
China 40 38 3 7 11
Hong Kong 17 12 2 0 1
Japan 29 24 3 2 3
Malaysia 27 36 1 2 7
Singapore 40 31 2 4 5
South Korea 36 33 3 8 7
Taiwan 13 10 0 1 4
Total 224 200 17 26 39
Table 9 - Tax Treaty Partners by Geographic Region
With respect to the overall use of anti-abuse rules in the bilateral tax treaties, the analysis of the tax
treaty networks of the eight jurisdictions shows a rather heterogeneous impression (see Table 10
below for an overview). The percentage of tax treaties that include anti-abuse rules of any sort
ranges from 24.24% (China – 24 tax treaties) to 84.38% (Hong Kong – 27 tax treaties) and is in total
291 More than 100 jurisdictions, including all of the eight Asia Pacific jurisdictions except for Taiwan studied in this paper, have committed to fast tracking the incorporation and implementation of certain minimum standard BEPS recommendation into existing tax treaties. On 24 November 2016, the OECD released The Multilateral Convention to Implement Tax Treaty Related Measures to Prevent BEPS (the “Multilateral Instrument“ or “MLI“) which was the result of the BEPS Action 15 work and, according to the OECD press release accompanying the Multilateral Instrument, will “implement minimum standards to counter treaty abuse and to improve dispute resolution mechanisms while providing flexibility to accommodate specific tax treaty policies.“ The text and the related explanatory statement have been formally adopted by approximately 100 countries. According to the official OECD information brochure accompanying the release of the MLI, MLI is “intended to transpose results from the OECD/G20 BEPS Project into more than 2,000 [existing] treaties worldwide,” With the first high-level signing ceremony expected to take place in early June 2017, in relation to BEPS Action 6 and the issue of whether to adopt PPT or LOB (or a combination of both) in their respective treaties, it is expected that countries will in a relatively short period of time determine which BEPS Action 6 recommended approach to adopt and register their decision pursuant to the MLI with the OECD.
48
43.48% (220 tax treaties). The absolute numbers have Singapore in first place with 48 (out of 82) tax
treaties including an anti-abuse rule of any sort, followed by South Korea with 35 (out of 87) and
Japan with 30 (out of 61).
While the use of anti-abuse rules in general is not comprehensive and highly diverse, the choice of
the preferred measure if an anti-abuse rule is incorporated into the treaty is quite homogeneous.
The one measure most often used is clearly the main purpose test, with 113 individual treaties
containing the test. This represents 22.33% of all treaties and 34.66% of all anti-abuse measures in
the treaties.
Jurisdiction Number of
Tax Treaties
LoB-
Clause MPT LoR Other None
Australia 44 2 10 9 5 27
China 99 4 19 1 21 75
Hong Kong 32 0 13 2 23 5
Japan 61 8 13 13 20 31
Malaysia 73 0 7 17 5 50
Singapore 82 0 18 39 8 34
South Korea 87 1 23 5 14 52
Taiwan 28 1 10 3 12 12
Total 506 16 113 89 108 286
Table 10 - Tax Treaties and Anti-Abuse Rules
The Limitation on Benefits clause on the other hand is the least applied anti-abuse rule with only 16
LoB clauses in all 506 treaties reviewed. Thus only 3.16% of the treaties include an LoB clause. Japan
has the most (8) LoB clauses in its treaties, while three jurisdictions (Hong Kong, Malaysia and
Singapore) do not have a single LoB clause in their current treaty network.
As mentioned above, the jurisdictions are somewhat diverse in their anti-abuse policy. From the
Australian tax treaties, one could conclude that anti-treaty shopping measures are not one of the
main priorities of the overall Australian tax treaty policy. More than 60% of Australia’s treaties do not
have an anti-abuse rule. Within the 17 treaties that include an anti-abuse measure of any sort, the
“main purpose test” is the one most utilized. When including anti-abuse measures in the treaty
though the Australian tax treaty policy seems to not only rely on one measure but often times
combines a number of different, sometimes even overlapping provisions.
Anti-treaty shopping measures seem to be of more importance to the Japanese tax treaty
negotiators as slightly more than 50% of Japan’s existing treaties have at least one anti-abuse rule.
The treaties, which have been negotiated and signed more recently all have some sort of an anti-
abuse rule. Within the 31 treaties that include an anti-abuse measure, the “main purpose test”,
subject-to-tax clauses and a general anti-abuse clause committing the contracting states are the
49
measures most utilized. Notably, almost every tax treaty that includes an anti-abuse measure usually
contains a number of different, sometimes even overlapping anti-abuse provisions. Even though
Japan has the most LoB clauses in its tax treaties as compared to the other seven jurisdictions,
however within the Japanese treaty network the LoB-clause is still the least utilized anti-abuse rule
within the existing Japanese treaty network.
Based on the review of the existing treaties, the South Korean tax treaty policy has a strong focus on
negotiating main purpose test provisions as the preferred anti-treaty shopping measure. South Korea
has the highest absolute number of MPT provisions in its tax treaties (23) and also the highest
relative share of MPT among the anti-abuse rules availed of (53.49%). While, most of the specific
treaty provisions are applicable to passive income, capital gains and other income, deviations from
this practice are also quite numerous with either extended or limited scopes for the MPT provision
found in certain treaties. The second most important anti-abuse measure is a specific anti-treaty
shopping rule that denies treaty benefits to intermediaries held/controlled by non-residents of either
contracting states.
China has, at least in the last 10 years, developed a very clear tax treaty policy with regards to anti-
treaty shopping. China strongly prefers the main purpose test, especially when (re-)negotiating tax
treaties with OECD member states. The MPT provisions in China’s tax treaty network are usually
applicable only to passive income (and other income in certain cases). Usually the specific anti-treaty
shopping rules are supplemented by a more general rule which declares that domestic general anti-
avoidance rule (GAAR) a priori as not treaty overriding. However not only the numerous application
of the main purpose test in the most recent tax treaties shows the preference of Chinese tax policy
for the test, also the 2008 introduced and since then constantly evolving general anti abuse rule of
the Chinese domestic income tax act.292 The Chinese GAAR also includes a purpose test, which denies
claimed tax benefits for transactions/arrangements that do not have “a reasonable business
purpose”293. As with all other principal or main purpose tests, the Chinese GAAR focuses on the
“main purpose” of the transaction – however, guidance as to how this main purpose can be
identified is very limited.294
292 See J. van der Pas, “Improving the Chinese General Anti-Avoidance rule: A Comparative and Functional Approach”, 8 World Tax Journal 1, 79-120.
293 See J. van der Pas, “Improving the Chinese General Anti-Avoidance rule: A Comparative and Functional Approach”, 8 World Tax Journal 1, 79-120.
294 See J. van der Pas, “Improving the Chinese General Anti-Avoidance rule: A Comparative and Functional Approach”, 8 World Tax Journal 1, 79-120.
50
Tax avoidance and treaty shopping seems to be a high priority agenda from Hong Kong’s tax treaty
policy perspective. Only 5 of the 32 tax treaties (15%) do not contain any anti-abuse rules. While the
LoB clause is never used, 13 treaties (40.63%) contain a MPT provision.
Malaysia’s tax treaty policy, at least until now, seems to be not overly concerned with treaty abuse
and treaty shopping as more than two thirds (68.49%) of its treaties do not contain any anti-abuse
provisions. Among the employed anti-abuse measures the MPT is only ranked second (7 treaties)
behind the Limitation of Relief provision (17 treaties). Compared to the LoB clause (zero applications)
the main purpose test seems to be the preferred choice of Malaysia.
Similar to Malaysia, Singapore’s tax treaty policy seems to prioritize concluding Limitation of Relief
provisions (39 treaties) over MPT provisions (18 treaties). However, there is a noticeable trend of the
MPT provision increasingly being added to Singapore’s treaties, especially the more recently signed
treaties (both new and renegotiated treaties). Almost 60% of Singapore’s treaties contain an anti-
abuse rule. The LoB clause however has not been adopted in any of Singapore’s existing treaties.
Taiwan which has the smallest tax treaty network of the eight jurisdictions studied in this paper has
opted for the main purpose test provision as its anti-treaty shopping measure of choice (10 of the 16
treaties with an anti-abuse rule contains a MPT provision). None of its existing treaties contains a
Limitation on Benefits provision.
The overarching historical trend of all eight jurisdictions shows a strong and persistent increase of
the usage of Anti-Avoidance measures of any sort since the year 2000 (see Figure 1). The total
number of Anti-Avoidance measures almost doubled from 1999 to 2009 and increased by 126%
between 2009 and 2016. Every new Treaty negotiated after 2009 contains on average 1.07 Anti-
Avoidance rules. Over the whole observation period 66% of Treaties contain an Anti-Avoidance
measure of any sort.
Figure 1 - Anti-Avoidance Measures per Treaty - Total
0.00
0.20
0.40
0.60
0.80
1.00
1.20
before 1990(first Treaty in 1946)
1990-1999 2000-2009 after 2009
Anti-Avoidance Measures per Treaty
Anti-Avoidance Measures per Treaty MPT per Treaty LoB per Treaty
51
Figure 1 also shows the persistent increase of MPTs as the most preferred Anti-Avoidance measure.
Over the whole observation period MPTs are implemented in 22% of all 506 Treaties analyzed. The
growth rate of MPTs in newly negotiated Treaties however is steadily increasing (3% before 1990, 8%
1990-1999, 19% 2000-2009, and 43% after 2009). Figure 1 also depicts the relative unimportance of
LoB-clauses very impressively.
This relative unimportance can also be seen from Figure 2, which relates the number of MPTs and
LoBs to the total number of Anti-Avoidance measures. While the LoB had some increased popularity
during the early 2000s (9% of new Anti-Avoidance measures were LoBs), this trend seems to
disappear (4% of new Anti-Avoidance measures were LoBs after 2009). In total LoB-clauses account
for 5% of Anti-Avoidance measures and are implemented in 3% of all Tax Treaties analyzed.
Figure 2 - MPT and LoB per Anti-Avoidance Measure
Already before 1990, the most preferred Anti-Avoidance measure is MPT (10% of all Anti-Avoidance
measures). This rate steadily increases (25%: 1990-1999, 31%: 2000-2009) as 40% of all new Anti-
Avoidance measures after 2009 are MPTs. Over the whole observation period 34% of Anti-Avoidance
measures are MPTs. And they are implemented in 22% of the Tax Treaties investigated.
The analysis shows clearly that for the eight selected Asian jurisdictions, between the PPT and the
LOB provisions recommended by the BEPS Action 6, the PPT has been the overwhelmingly preferred
anti-treaty shopping provision in the existing treaties. While the criticism in the literature and also in
the discussion of this measure above might be justified because of the subjective nature of the PPT
and the attached difficulties with respect to proving intentions and motives, the eight jurisdictions
are very likely to opt for the PPT provision in the post-BEPS treaty world. The only exception may be
Japan which, depending on the other contracting state involved, could opt for either the PPT or the
LoB provisions as the preferred choice. Over the next several years, many of the existing treaties are
0.00
0.10
0.20
0.30
0.40
0.50
before 1990(first Treaty in 1946)
1990-1999 2000-2009 after 2009
MPT and LoB per Anti-Avoidance Measure
MPT per Anti-Avoidance Measure LoB per Anti-Avoidance Measure
52
likely to be amended on a fast track basis to incorporate the BEPS Action 6 recommendations under
the OECD BEPS Multilateral Instrument and many of the new treaties will explicitly contain either the
PPT or the LoB provisions (or a combination of both). Also, the likely shift of burden of proof under
the PPT scrutiny towards the taxpayer is a legitimate concern against this measure, especially in
jurisdictions where clear tax law guidance is lacking. It is of vital importance for taxpayers to
anticipate the application of PPT scrutiny as a condition precedent to a treaty claim to their existing
and proposed transactions.
On a brighter note, compared to the LoB clause with its various highly complex and interrelated
tests, exceptions and counter-exceptions, the main purpose test might be simpler, more flexible,
easier to apply and less difficult to agree upon the specific wording during tax treaty negotiations.
Additionally, the Asia Pacific jurisdictions analyzed in this paper arguably have familiarity and
experience with the PPT given the prevalence in existing treaties.